Agile Software Architectural Governance in the Real World

Abstract
This article focuses on the use of the ‘Architect’ role within Agile environments by taking into consideration the experience of the author, as well as objective opinions from other Software professionals who have found their own version of successful software architecture via different means in the agile environment.  The consideration of using the ‘architect’ role within an agile team is discussed along with how the architect can ensure the software does not become fragmented as well as ensuring architectural governance and accountability.

 

Architecture for any software project is the glue and foundation that keeps the software together, keeps it stable, keeps it maintainable, and keeps it performing well, among other things. How you “get to the architecture” depends on many factors, including, the structure of the project, the methodology used, the people on the project, and other factors. This article is going to focus on architecture as it is used specifically within Agile environments.

In addition to my own background and insight having worked in Architecture in Agile (as well as waterfall and others), recently, I put out a few questions to the architecture and agile communities hoping to gain additional opinions from those working with Architecture in Agile environments and how architecture is used and influenced in these different cultures. The response and encouragement has been overwhelming. Because this article is part opinion and part research, I take into account the experiences and opinions of myself as well as other top professionals. My goal is to be as objective as possible while sharing the experiences and opinions of other professionals weather or not I have shared similar experiences, use different methodologies, or agree or disagree.

Using the waterfall methodology, many software projects try to nail down all of the requirements up front, and these include both functional and non-functional requirements. Architecture documents are also created to various levels of detail describing the architecture. Waterfall typically can’t account well for variations in business or technical requirements along the way, so trying to get it right the first time (never happens) is typically important for many organizations. So typically, lots of time is spent up front tying to nail this down as best as possible.

Agile teams handle architecture differently, and depending on the team, there may be an initial iteration (or two, or three, etc) to nail down the initial architecture. Some agile teams will try to at least nail the most significant decisions as related to architecture (see, Introducing Significant Architectural Change within the Agile Iterative Development Process) hopefully trying to mitigate future architectural changes while understanding the costs associated with it. Other Agile teams let their software grow organically, as many Agile proponents promote YAGNI (You ain’t gonna need it), or also verbosely explained as “let’s not write anything until we actually have a business or technical reason to do it – aka, let’s not over-architect). So, YAGNI is great for a lot of things, but it doesn’t go far enough to account for significant architectural decisions that need to be baked into the software without significant cost down the road. The “Last Responsible Moment” principal tries to address this as well, and it works great for some teams, but it’s up to teams to determine when the “Last Responsible Moment” for creating and implementing the architecture is, and that becomes very subjective with the prospect that if you wait too long, the more re-engineering work is required.

So, how are Agile teams doing Architecture out in the field? My experience tells me how I’ve done it, as well as what has worked in the past for me and my teams. How does everybody else do it? Do they do it like me, do they have a better way, do they do it worse? Are they successful? Knowing that “the team” makes all the difference and what works for one team won’t necessarily work for another team, I have compiled what I think are the best responses, along with my commentary, to the questions put out to the community.

Dani Mannes – Agile Modellers & Developers

Dani Mannes is the Founder and Chief Architect at ACTL Systems Ltd. His work focuses on the defence industry where he is as a consultant and trainer to helping this predominately waterfall industry adopt agile.

In Dani’s approach, he uses the terms “agile developers” and “agile modellers” to define agile team members which do either development or design. He runs into a typical problem that I have seen in many Agile environments, and as Dani puts it “The teams are supposed to refactor, but they often don’t do it because of time pressure. This leads ultimately to spaghetti code/architecture and the velocity will eventually drop dramatically.”

The modellers will use a modelling tool to sketch out and ensure architecture documentation is up to date. “So in each sprint you have an architecture description of the sprint scope. But the architecture should not only focus on the current sprint but also take into consideration stories that will most probably be tackled in the next 2 sprints”. Dani emphasizes that taking into consideration the future stories is essential, but these should not be modelled at this point since “taking into consideration means only to think about them but not actually find a solution for them”.

In Dani’s world, ”the team acts as the architect”, and he states that there is no need for the architect role. But, he does keep one person in the role of architect, or “architect champion” just to keep discussions short, and to monitor the need for architectural change during each sprint.

“Our experience has shown that when the team applies a model based design process during first days of each sprint where focus is set on the sprint scope and attention is given to the scope of the next 2-3 sprints, the team is capable of coming up with a good architecture that serves as guidance during implementing the sprint scope.”, says Dani, and “Since the team has come up together with the architecture, all members are aware of the modules”.

Lee Fox – Architecture as Part of The Team

Lee Fox is a Software and Cloud Architect, Agilist, and Innovator, and he ensures that the architect is always a contributor and a team player. In his experience with waterfall projects, the architect is isolated from the rest of the team and isn’t necessarily even part of it, “As part of the team, I preach that the architect MUST be a contributing member and with some degree of consistency even contribute to team deliverables.”, and “the architect needs to really enhance the idea of empowerment and encourage the team to make architectural decisions.” Lee Fox also values that the architect must have the teams trust and maintain the big picture vision. This is a recurring theme in many agile processes when it comes to working within an agile team.

Lee’s vision is that “Agile architects work both in the low level with the team as well as the high level with the business. They use their broad exposure with the business to help guide a team’s decisions in the right direction. “ He is fine with architects working on multiple teams, but cautions that the “the architect must contribute to each team he is a member of and keep up with the big picture”. What Lee has seen through his approach and coaching is an increase in both velocity and code quality.

The Need for Governance – Dan’s Thoughts

In Agile, teams are self organized where there is no one on the team who should have additional responsibilities than anyone else as everyone is working together to achieve the same goal. Work is picked up and worked on by any team member and the expertise that is created is shared amongst all team members. However, I believe, as well as other respected Software Professionals, such as, Simon Brown, author of Software Architecture for Developers (https://leanpub.com/software-architecture-for-developers), that you need someone responsible for the Big Picture of the Software, and that includes the architecture. This creates accountability and governance for the architecture and all non-functional requirements (NFRs).

There are multiple ways to approach this, but ultimately realizing that having that architect who is responsible for the big picture can help ensure that the architecture is continually in-line with the functional and non-functional requirements. The architect will have increased access to both the technical side and the business side to ensure that the team(s) are continually aligning to not only the business/functional requirements, but the architecture is in alignment with both short and long term functional and non functional requirements, and that the existing architecture is followed, revised, and re-worked as necessary.

Ok, so I know some agile purists out there are thinking “Long term requirements are very subjective. Until a user story gets chosen by a ‘Product Manager’ and moves from the backlog to being actively worked on during a sprint, it’s not really a requirement yet”. Ok fine. I get that, and I understand the advantages here which is makes agile a process which helps you to change direction or add new to the project features midway through a development phase. User stories are (should be) always functional requirements though. When considering the architecture, we need to understand what is in the backlog or the general type of functional items that are in the backlog which helps the architect to create the technical vision. The technical vision can change as the project progresses, but ultimately understanding the grand vision will help ensure an architecture that takes into account current requirements and ensure it meets future product functionality with minimal re-work.

A drawback of Agile teams without architectural governance is that the system tends to fragment or suffers from too much rework and often non-functional requirements (such as performance, scalability, and others as related to architecture) get tossed out the window. Imagine a team has estimated a total of 20 points for the upcoming sprint. They have to consider, how can they get it done? Among these considerations is what is needed now, and unfortunately what is needed now, is often at the expense of technical accountability. Fragmentation of the system (or of the architecture) occurs when the now becomes the most important piece rather than ensuring we are adhering to sound architectural principals an meeting our NFRs as well as the long term architectural vision of the software. This is why architectural governance is important.

In agile planning meetings, the team will typically talk about design and may also discuss architectural changes. Ultimately, the team may still be on their own to make these decisions and ensure that the architecture they have will meet the existing business and technical requirements of the sprint, but it’s up to the architect to ensure that the decisions and approaches undertaken by the team are in fact in line and consistent with the architectural vision. The architect is accountable, and accountable for the team’s decisions related to architecture and any architectural changes that come about.

The architect is also a team member who may also code but ultimately has responsibility for the continuous evolvement of the architecture. Typically the architect needs to work in the trenches alongside the development and business teams to ensure constant technical team and business communication. The architect should be involved in all of the agile processes from planning, development, to retrospectives. If the architecture has failed, or we spent too much time on a sprint worrying about “the now” and compromised our architecture, it needs to be brought up and addressed (think retrospective). The whole team should be able to come up with reasons why it failed and how we can improve.

In Summary

When using agile teams, the need for the role of the architect needs to have a high level of consideration. It’s important for this role to work alongside the team and have a very sound vision of how the architecture will meet short and long term objectives with minimal re-work. The architect should be part of all regular agile processes and held to accountability while allowing the entire team to propose and come up with architecture, but ensuring the architect is accountable. Having the architect maintain accountability provides a level of governance to the project from a technical architecture perspective that would otherwise get lost and lead to fragmentation in most agile environments.

Advertisements

Are Scrum Processes Such As Burndowns and Task Estimation Working Against Your Organization’s Agility?

Abstract

This article contrasts Scrum with Agile methodologies and points out that Scrum can fit within an Agile development team, but Scrum alone doesn’t mean your organization has become Agile.  Scrum processes such as task estimation and burndowns may be established Scrum processes, but upon further analyzing their effects you see that they could actually be working against your team’s agility.  It’s important to evaluate the value of established processes, to determine if they are helping or are detrimental to your organization’s Agile initiatives.  A real world example is used to demonstrate an example of how a good intentioned process becomes inefficient within an organization’s Agile implementation.


Are Scrum Processes Such As Burndowns and Task Estimation Working Against Your Organization’s Agility?

Recently, a colleague asked me about task estimation and how to reduce the amount of time it’s taking their team to do it.  I gave him some suggestions, but it got me thinking more about task estimation and why agile teams are doing it at all.  I’ve also been involved with many organizations who are following Scrum processes, calling themselves agile, and they are getting there, but they are still putting Scrum processes before people and not really thinking about the value of the processes they are following.  This article is going to talk about how using Scrum processes such as burndown charts and task estimation can actually be working against your organization’s initiatives to improving agility.

Scrum actually predates agile, so it’s fair to say that just because you are using Scrum you are not necessarily embracing agile practices, however when you look up Scrum in Wikipedia you will see the following “Scrum is an iterative and incremental agile software development framework for managing software projects and product or application development.”.  Ok, so it is an “agile software development framework”, so you can forgive people’s perception (even mine) when they feel that by following scrum processes we are Agile.  But, I believe there still needs to be a further distinction here.  Scrum is still just a process for managing software development, and I see Scrum as a set of rules that “could” be used in an agile environment, but by following Scrum processes to the letter of the law, we could actually be putting processes over people – which is definitely the opposite of what the agile manifesto is trying to achieve.

Scrum, like many other agile implementations has the basic premise of a user story backlog.  We estimate story points with a number of relativity, instead of using an absolute measurement of time, and over time we can determine our velocity – which is how many story points we can generally complete in a sprint or iteration.  We can now easily judge our backlog to get an idea of how long it will take to complete the stories in the backlog.

Once estimated, we task out the stories for the sprint, but unlike standard Scrum implementations, I don’t like to estimate hours for tasks.  Inherently there is nothing wrong with it, but myself and many Agile experts don’t see a lot of value .  It’s just very difficult to estimate work accurately in absolute measurements of time – that’s why we story point the stories to begin with.  Estimating at even the task level also has the same problems of not being so accurate.

With Scrum implementations, estimating tasks in absolute time is still a vital part of the process.  Wikipedia defines a scrum task as the following: “Added to the story at the beginning of a sprint and broken down into hours. Each task should not exceed 12 hours, but it’s common for teams to insist that a task take no more than a day to finish. “ [Author’s note, the latest version of the Scrum guide has removed task estimation from the Scrum requirements]

I had a friendly discussion with a previous client on estimating tasks and I could never get a good answer as to why they do it other than “it’s part of the process”, and “it’s agile”.  There certainly is a lot of confusion out there about what exactly is Agile, and what constitutes adding value as part of an agile implementation.  Sometimes teams feel that by following a Scrum process, we’re agile – but as a team we need to think about value.  Scrum, agile, or not – what value are we actually getting from spending the time to estimate tasks?  If, as a team, we cannot answer that question we need to re-evaluate this process and determine if it’s a waste of time or not.  If there is value, sure let’s continue doing it, but many times a process is followed for process sake.

However, it’s easy to see why task estimation is being done.  We need the information for our burndown charts.  These charts tells us how many task hours we have completed versus how many hours are remaining and are part of the Scrum process that should be presented during the daily stand-up meeting.  Ok, sure – then theoretically, IF we need burndown charts, then we need tasks estimated in hours.

But, do we need burndown charts?

There are a few teams out there that can accurately estimate blocks of development work in absolute units of time, but it’s not the majority.  The reason we estimate in story points for user stories is because software estimates in absolute measurements of time are barely ever accurate.  So, why does Scrum insist on estimating in story points for user stories, but insist on estimating task hours for individual tasks within the user stories?  Even when estimating dozens of small tasks individually we succumb to the same inaccuracy as we would if we were estimating user stories the same way.

The only thing that is important at the end of the sprint (or iteration) is a completed story.  A non-completed story isn’t worth anything at the end of the sprint even if 10 of 12 hours are completed.  A quick look at completed tasks vs non-completed tasks should be enough motivation for the team to know and decide what needs to be done to complete the work by the end of the sprint.  Many agile experts would agree, including George Dinwiddie who recommends in a Better Software article to use other indicators instead of hours remaining for the burndown charts such as burning down or counting story points.  Gil Broza, the author of “The Human Side of Agile” will also recommend that burndown charts shouldn’t be used at all, and instead among other things, use swim lanes for tracking progress.

In my experience, knowing the number of hours outstanding in a sprint by looking at task hours doesn’t help and is an inaccurate metric to use to plan additional “resource” hours in the sprint.  Even though some organizations do this, the thought of using task hours to help plan for additional “resource” hours during a sprint isn’t valuable since the absolute time measurements aren’t accurate.

If you really are determined to be Agile, you need to make sure you understand where processes make sense,  After all, the agile manifesto preaches “individuals and interactions over processes and tools”.  Following Scrum processes doesn’t necessarily make you Agile, and you need to really think about the value of these processes and determine if they are working for or against your initiatives to become more agile.  In the real world examples of task estimation and burndown charts, it was clear that these established Scrum processes were actually working against the organization in terms of becoming more agile.  Even if your goal isn’t to be agile, by identifying and eliminating processes that don’t add value, you are eliminating waste and opening the door to replace the processes with initiatives that can truly add new value.

Dan Douglas is a professional independent Software Consultant and an experienced and proven subject matter expert, decision maker, and leader in the area of Software Architecture. His professional experience represents over 12 years of architecting and developing highly successful large scale solutions.  Dan has been the architect lead on over 15 development projects.

A Little Documentation Please, Defining Just Enough Code Documentation for Agile Manifesto No. 2

Abstract

This article helps clarify agile manifesto number two “Working software over comprehensive documentation” by defining the different types of documentation being referred to. An analysis is given on code documentation indicating the complexities that warrant it, and how refactoring and pair programming can also help to reduce complexity. Finally, an approach is given on how to evaluate when to document your code and when not to.


A Little Documentation Please, Defining Just Enough Code Documentation for Agile Manifesto No. 2

The second item in the agile manifesto “Working software over comprehensive documentation” indicates that working software is more valued than comprehensive documentation, but it’s important to note that there is still some real value in documentation.

I see two types of documentation being referred to here. 1) Code and technical documentation typically created in the code by developers working on the system, and 2) System and architectural documentation created by a combination of developers and architects to document the system at a higher level.

I will save discussing system and architectural documentation for future articles, as it is a much more in-depth topic.

So let’s discuss the first point – code documentation

Questions that many teams ask (agile, or not) are – “How much code documentation is enough?” or “Do we need to document our code at all?”.  Some teams don’t ask, and subsequently don’t document.  Many TDD and Agile Experts will indicate that TDD and will go a long way to self-document your code.  You don’t necessarily need to do TDD, but there is a general consensus that good code should be somewhat self-documenting – to what level is subjective and opinions will vary.

Software code can be self-documenting, but there are almost always complex use cases or business logic that need to have more thorough documentation. In these cases it’s important that just enough documentation is created to reduce future maintenance and technical debt cost.  Documentation in the code helps people understand and visualize what the code is supposed to do.

If technology is complex to implement, does something that is unexpected such as indirectly affecting another part of the system, or is difficult to understand without examining every step (or line of code), then you should at a minimum consider refactoring the code to make it easier to follow and understand.  Refactoring may only go so far, and there may be complexities that need to be documented even if the code has been nicely refactored.  If the code cannot be refactored for any reason, you have another warrant for documentation.

Think about what can happen if code isn’t documented well.  Developers will spend too much time looking at complex code trying to figure out what a system does, how it works, and how it affects other systems.  A developer may also jump in without a full understanding of how everything works and what the effects are on other systems.  This could create serious regression bugs and create technical debt if the original intent of the code is deviated from.  A little documentation gives you the quick facts about the intent of the code.  Something every developer will value when looking at the code in the future.

Of course, in addition to documentation, pair programming can and should be used to aid in knowledge transfer and can be a good mechanism for peers to help each other understand how the code and system work.  Pair programming is also a good mechanism for helping junior and intermediate developers understand when and where they should be documenting their code.

The way I distinguish what code should be documented and what shouldn’t be is based on future maintenance cost.  Consider how your documentation lends itself to ease of ongoing maintenance and how easing the ongoing maintenance will reduce technical debt and contribute to working software over time.  If your documentation will directly contribute to working software by eliminating future complexity and maintenance, then document. If there is no value to the ongoing future maintenance, then don’t.  If you are unsure, ask someone a little more senior, or do some pair programming to help figure it out.  I’ve also seen value in peer reviews to help ensure documentation is being covered adequately, but I still prefer to instill trust within the team to get it done properly rather than a formal review process.  When in doubt, document – a little more documentation is always better than missing documentation.

Note: My motivation for writing this article came after reading a good article on the same topic http://www.boost.co.nz/blog/random-thoughts/agile-manifesto-number-two/

A question came in the comments on how to weight the amount of documentation needed in a project.  My article was meant to address that question regarding code documentation with my own opinions.  System level documentation is much more in-depth, so look for a future article addressing documentation at the system level.

-Dan

Dan Douglas is a professional independent Software Consultant and an experienced and proven subject matter expert, decision maker, and leader in the area of Software Architecture. His professional experience represents over 12 years of architecting and developing highly successful large scale solutions.  Dan has been the architect lead on over 15 development projects.

Navigating the Technology Stack to Get a Bigger Return on Your Technology Selection

Abstract

This article introduces the reasons why organizations choose to standardize a technology stack to use on existing and future projects in order to maximize ROI of their technology choices. When selecting technology for new projects, the architect should consider both technologies within and outside of the existing technology stack, but the big picture needs to be carefully understood and consideration needs to be placed on the ROI of introducing new technology versus using existing technology within the technology stack.


Navigating the Technology Stack to Get a Bigger Return on Your Technology Selection

As a Software Architect, understanding the long term affects and ROI of technology selection is critical.  When thinking about technology selection for a new and upcoming project, you need to consider the project requirements, but also look beyond them at the bigger picture.  Even though at first glance, a specific technology might seem perfect for a new project, there may already be a more familiar technology that will actually have a much bigger return on investment in the long term.

Many organizations stick to specific technology stacks to avoid the cost, overhead, and complexity of dealing with too many platforms.   An architect should have specialized technical knowledge of the technology stack used by the organization, and if the organization’s technology stack isn’t standardized, the architect should work to standardize it.

Advantages to an organization by standardizing their technology stack

Development costs – It’s easier to find employees who are specialized on a specific platform instead of having multiple platform specializations with conflicting technologies.  When you need developers with conflicting skillsets (ex: .NET and Java), you will likely need to pay for additional employees to fill the specialization gap.

Licensing costs – It’s typically advantageous to stick with only a few technology vendors to attain better group discounts and lower licensing costs.

Physical tier costs – It’s cheaper and more manageable to manage physical tiers that use the same platforms or technologies.  Using multiple platforms (ex: Apache and Windows based web servers require double the skillset to maintain both server types and to develop applications that work in both environments)

Understand the bigger picture beyond your project to better understand ROI

As an architect you have responsibility with technology selection as it retains to ROI.  Once you are familiar with the organization’s technology stack and the related constraints, you can make a better decision related to technology selection for your new project.  You may want to put a higher weight on technology choices known collectively by your team, but it comes down to understanding the bigger picture beyond your current project and understanding the ROI of both the project and ongoing costs of the technology choices used.  You may need to deviate from your existing technology stack to get a bigger ROI, but be careful that the cost of supporting multiple platforms and technologies doesn’t exceed the cost savings of using a specific specialized technology for a specific case in the long term.

When Microsoft released .NET and related languages (VB.NET and C#) in 2001, many organizations made the choice to adopt VB.NET or C# and fade out classic VB development.  Those that made the switch early had an initial learning curve cost.  Organization’s that chose to keep their development technology choice as classic VB eliminated additional costs at the onset, however they paid a bigger price later when employees left the company, finding classic VB developers became more difficult, and the technology became so out of date that maintenance costs and technical debt began to increase dramatically.

Sometimes the choice and ROI will be obvious; the technology in question might not be in use by your organization, but it lends itself well to your existing technology.  For example, introducing SQL Server Reporting Services is a logical next step if you are already using SQL Server, or introducing WPF and WCF will compliment an organization that is already familiar with development on the Microsoft .NET platform.

In another case, it may make sense to add completely new technology to your technology stack.  For example, it may be advantageous from a cost perspective to roll out Apple iPhones and iPads to your users in the field, even though your primary development technology has been Microsoft based.  Users are already familiar with the devices, and there are many existing productivity apps they will benefit from.  Developing mobile applications will require an investment to learn Apple iOS development or HTML5 development, but the total ROI will be higher than if the organization decided to roll out Microsoft Windows 8 based devices just because their development team is more familiar with Windows platform development.

Finally, there will be cases where even though the new technology solves a business problem more elegantly than your existing technology stack could, it doesn’t make sense to do a complete platform change in order to get there.  In these cases, the ongoing licensing costs, costs of hiring specialized people, and complexities introduced down the line far outweigh any benefits gained by using the new technology.

Summary

It’s important that the software architect facilitates the technology selection process by evaluating technology based on ROI of the project while also considering the long term ROI and associated costs of the selected technology.  It’s important not to focus only on your existing technology stack however, and consideration should be given to unknown or emerging technologies within the technology selection process.  Careful consideration should be given to the cost of change and ongoing maintenance of any new technology and the ROI needs to be evaluated against the ROI of sticking to the existing technology stack over the long term.

Dan Douglas is a professional independent Software Consultant and an experienced and proven subject matter expert, decision maker, and leader in the area of Software Architecture. His professional experience represents over 12 years of architecting and developing highly successful large scale solutions.  Dan has been the architect lead on over 15 development projects.

Introducing Significant Architectural Change within the Agile Iterative Development Process

Abstract

As an architect working within an iterative agile environment, it’s important to understand how significant architectural decisions need to be made as early in the development process as possible to mitigate the high cost of change of making these decisions too late in the game. In iterative development, it’s important to realize the distinction between requirements that require significant consideration to the architecture and those that are insignificant. This article contrasts the difference between significant and insignificant requirements and demonstrates the best approach to implementing both types of requirements. In agile environments, it’s important to realize that significant architecture decisions will sometimes need to be determined iteratively and late in the development cycle. Guidance is provided on determining how to move forward by considering many factors including ROI, risk, cost of change, scope, alternative options, regression, testing bottlenecks, release dates, and more. Further guidance is provided on how the architect can collaboratively move forward with an approach to implementation while ensuring team vision and architectural alignment with the business requirement.


Introducing Significant Architectural Change within the Agile Iterative Development Process

As agile methodologies such as scrum and XP focus on iterative development (that is, development completed within short iterations of days or weeks), it’s important to distinguish the requirements that are significant to your architecture within the iterative development process.  Iteratively contributing to your software architecture is very important to maintaining your architecture, ensuring the right balance between architecture and over architecture, and ensuring that the architecture is aligned with the business objectives and ROI throughout the iterative development process

Many agile teams are not making a conscious effort to ensure significant decisions relating to software architecture are being accounted for iteratively throughout the software development process.  Sometimes it’s because there is a rush to complete features as required within the iteration and little thought is given to significant architectural changes, or sometimes it is lack of experience or lack of a team vision.  There is usually a lack of understanding how the important guiding principles of the architecture need to be continually established and how they shape the finished product and align with the business objectives in order to see a return on investment. This is where the architect plays a huge role within the iterative development process.

The lack of attentiveness to significant architecture decisions whether at the beginning or mid-way through a release cycle can cause significant long term costs and delay product shipping significantly. Many teams are finding out too late in the game that guiding principles of their architecture are not in place and strategies to get to that point still need to be put in place at the expense of time, technical debt, and being late to ship.  When requirements from the product team involve core architecture changes or re-engineering, changes are sometimes done so without recognizing the need to strategize and ensure that guiding principles of your core-architecture are in place to ensure on-going business alignment and minimal technical debt cost.

Within the iterative development process, it is important that the agile development teams (including the architect) learn to recognize when new requirements are significant and when they are not.  Deciding which requirements are significant and will be carried on as guiding principles of your architecture can be worked out collaboratively with the developers and architects during iteration planning or during a collaborative architectural review session.  This will help ensure that development is not started without consideration to the architecture of these new significant requirements, and that there is time to get team buy-in, ensure business alignment, and create a shared vision of the new guiding principles of your architecture.

In addition, to help prevent surprises during iteration planning, the architect can be involved and work with the product team when preparing the user story backlog to help identify stories that could have a significant impact on the architecture prior to iteration planning.  Steps can be put in place to help the product team understand the impact and to assist the product team in understanding what the ROI should be in contrast to the cost to implement major architectural changes.

So, what distinguishes significant architectural requirements with insignificant ones?

Separate which requirements have architectural significance with ones that do not where significant is distinguished by the alignment of business objectives and a high cost to change, and insignificant is more closely aligned with changing functional requirements within the iterative development process.

An insignificant architectural requirement will be significant as it retains to the functional requirements’, but not significant in terms of the core architecture.  To further contrast the difference between which decisions to consider significant and which to consider insignificant take a look at the following table.

Significant Insignificant
A high cost to change later if we get it wrong We write code to satisfy functional requirements and can easily refactor as needed
Functionality is highly aligned to key business drivers such as modular components paid for by our customers and customer platform requirements A new functional requirement that can be improved or duplicated later by means of refactoring if necessary.
The impact is core to the application and introducing this functionality too late in the game will have a very high refactoring and technical debt cost Impact is localized and can be refactored easily in the future if necessary.
Decisions that affect the direction of development, including platform, database selection, technology selection, development tools, etc. Development decisions as to which design patterns to use, when to refactor existing components and decouple them for re-use, how to introduce new behavior and functionality, etc.
Some of the ‘ilities’ (incl: Scalability, maintainability/testability and portability). Some of the ‘ilities’, such as usability depending on the specific case can be better mapped to functional requirements.  Some functional requirements may require more usability engineering than others.

It is best to handle the significant decisions as early as possible in the development process.  As contrasted in the table below, you can see how the iterative approach lends itself well to requirements that have an insignificant impact on the architecture.  You can also see how significant architectural requirements really form guiding principles of your architecture and getting it right early on lessons the impact of change on the product.

Insignificant Decisions How to Approach
Examples:

  • New functional requirements (ex: to allow users   to export a report to PDF. There is talk in the future about allowing export to Excel as well, but it is currently not in scope.)
  • Modifications and additions to existing   functionality and business logic

 

Use an agile iterative approach to development.  This is a functional requirement, with a low cost to change and a low cost to refactoring.  Write the component to only handle its specific case and don’t plan your code too much for what you think the future requirements might be.  If the time comes to add or improve functionality, then we refactor the original code to expose a common interface, use repeatable patterns, etc, In true agile form, this will prevent over-architecture if future advances within the functional requirements are never realized, and there is minimal cost to refactor if  necessary.
Significant Decisions How to Approach
Examples:

  • Our customers are using both Oracle and SQL Server
  • Performance and scalability
  • Security Considerations
  • Core application features that have a profound   impact on the rest of the system (example, a shared undo/redo system across all existing components) 
These decisions need to be made early as possible and an architectural approach has to be put in place to satisfy the business and software requirements’.  These decisions are usually significant as there is a huge cost to change (refactoring and technical debt) and potential revenue loss if not put in place correctly, or needs to be refactored later.  These decisions are core to the key business requirements and will have a huge cost to change if refactoring is required later.

Agile Isn’t Optimized For Significant Architectural Change

It’s unfair to assume that the agile development process is built to excel at the introduction of significant functionality and architecture changes without a large cost.  This is why some requirements are significant and need to be made as early on in the development cycle as possible while ensuring there is alignment with the business objectives.

As great as it would be to map out every possible significant requirement as early on as possible, there are sometimes surprises.  This is agile development after all.  We need to understand that along with changing functional requirements, significant business changes part way through the development process can occur and could still have a significant impact on our core architecture and guiding principles, so we need to mitigate and strategize how to move forward.

Certainly, it’s possible to introduce significant core-architecture changes by means of refactoring, or scrapping old code and writing new functionality – that’s the agile approach to changing functional requirements and how agile can help prevent over-architecture and ensure we are only developing what’s needed.  The problem is that it doesn’t work well for the significant decisions of your architecture, and when we do refactor, the cost can be so high that it will exponentially increase your time to market, cause revenue or customer loss, and potential disruption to your business.  In these cases, the architect, along with the product and development team need to create a plan to get to where they need to be.

Mitigating the Cost of Significant Change

There is always a cost to introducing core architectural functionality too late in the game.  The higher the cost of the change and the higher the risk of impact, the more thought and consideration needs to be put into the points below.  Here are some points that need to be considered.

  • The refactoring cost will be high.  Is there an alternative way we can introduce this functionality in a way that will have a minimal impact now without affecting the integrity of the system later?
  • This change is significant and will require a huge burden by the development team to get it right. Will we have a significant ROI to justify the huge cost to change?  For example, is Oracle support really necessary after developing only for Sql Server for the first 6 months or is it just a wish from the product team?  Do we really have customers that will only buy our product if it’s Oracle, and what are the sales projections for these customers?  Is there a way we can convince these customers to buy a Sql Server version of the product?  The architect needs to work with the product and business teams to determine next steps.
  • How will this affect regression testing?  Are we creating a burden for the testing team that will require a massive regression testing initiative that will push back our ship date even further?  Is it worth it?
  • How close are we to release?  Do we have time to do this and make our release ship date?
  • What is the impact of delaying our product release because of this change?
  • Is it critical for this release or can it wait until a later release?
  • Can we compromise on functionality?  Can we introduce some functionality now to satisfy some of the core requirements and put a plan in place to slowly introduce refactoring and change to have a minimal impact up front, but still allow us to meet our goal in the future?
  • What is the minimal amount of work and refactoring we need to do?
  • What is the risk of regression bugs implementing these major changes late in the game?  Do we have capacity in our testing team to regression test while also testing new functionality?
  • Are we introducing unnecessary complexity?  Can we make this simple?

All individuals involved in the software need to be aware of the impact and high cost that significant late to game changes will have on the system, development and testing teams, ship dates, complexity, refactoring, and technical debt that could be introduced.  There are strategies that can be used, and the points above are a great start in determining how to strategize the implementation of significant architecture changes.  One of the roles of the architect is to help facilitate and create the architecture and guiding principles of the system and ensure its long term consistency.  As the system grows larger and more development is complete, introducing significant architecture changes becomes more complex.  The architect needs to work with all facets of the business (developers, qa, product team, sales, business teams, business analysts, executives, etc) to help ensure business alignment and a solid ROI of significant architecture decisions.

Moving Forward

Once a significant decision is made that will form part of your architectures significant guiding principles, the architect needs to understand the scope of work, determine what will be included and what won’t, collaboratively create a plan on how we will get there, and understand how the changes will fit within the iterative development cycle moving forward.  The architect needs to ensure that the product and development teams share the vision, understand the reasons we are introducing the significant change, and understand the work that will be required to get there.  If your team is not already actively pairing, it may be a good time to introduce it or alternatively introduce peer reviews or other mechanisms to help ensure consistent quality when refactoring existing code to support significant architectural changes.

Depending on the level of complexity, the testing team may need to adjust their testing process to ensure adequate regression testing takes place that tests new and existing requirements that are affected by the significant architecture change.  For example, if we make a significant change to support both Oracle and Sql Server we need to ensure existing functionality that was only tested for Sql Server support is now re-tested in both Oracle and Sql Server environments.  The architect or developers can work with the testing team for a short time to help determine the degree of testing and which pieces of functionality specifically need to be focused on and tested to ensure the QA/testing teams are correctly focusing their efforts.

Summary

It’s important to distinguish significant architecture decisions from requirements that are insignificant as they relate to the core architecture of your system.  When introducing significant architecture changes iteratively within an agile environment, it’s important to understand the impact and complexity that significant changes have when introduced late in the game.   It’s important to understand the business impact of the changes and it’s important that the architect works with the rest of the organization to determine business alignment, risk, and ROI of these changes while understanding the cost of change before moving forward with a plan to introduce the significant architectural changes.


Dan Douglas is a professional independent Software Consultant and an experienced and proven subject matter expert, decision maker, and leader in the area of Software Architecture. His professional experience represents over 12 years of architecting and developing highly successful large scale solutions.  Dan has been the architect lead on over 15 development projects.

Your Team’s Ongoing Vision Will Determine the Long Term Consistency and Alignment of Your Software Architecture

Abstract

The ongoing consistency of your software architecture depends on its alignment with your business objectives, how suitable the technology choices are, and that your team has bought into and is following the architectural vision. This article demonstrates key factors in determining how suitable your software architecture is in being able to sustain a collaborative long term vision and growth of your software along with the difficulties of trying to ensuring business value without a shared vision or collaborative team input. Real world examples from the field are used to demonstrate successful scenarios, the thought behind them, how buy-in was attained, and how they complemented the QA testing strategy of the business objectives. An approach is given on how to recover failing projects where there was no consistent strategy and to turn chaos into a coherent strategy that is aligned with the business objectives.


Your Team’s Ongoing Vision Will Determine the Long Term Consistency and Alignment of Your Software Architecture

I’ve written a recent article that focused on ensuring that your architecture was aligned with your business objectives.  See  Aligning Your Software Architecture With Your Key Business Objectives and Why Your Business Needs It.  This article goes a step further and demonstrates how team buy-in and a shared team vision contributes to the consistency of the architecture and business value throughout the development cycle.

One of the problems that the architecture of your software should address is consistency. Consistency with technology, design patterns, approaches, layers, frameworks, etc.  As software projects evolve, it’s important to ensure that the architecture, especially as aligned with your business objectives, remains consistent.  Functional requirements can certainly change along the way, but the key is to do enough architecture work, and ensure consistency, for those big design decisions that need to be made as early in the game as possible.

Ensuring consistency in this capacity isn’t about guidelines for best practices such as re-use of existing components and developing components that are decoupled from one another.  These best practices should be part of most (all) development projects.  Ensuring consistency of the architecture is about ensuring that the guiding principles of your core-architecture have buy-in, are clear, are being followed, and are driving the long term success of your software in relation to your critical business objectives.

Ensuring consistency can require a certain level of control in some scenarios, however more often than not, very little controls are required when you have team buy-in, and a shared vision from the beginning as to how the business value is being provided.  This is especially true once you have a team that is consistently delivering business value with the software.  Enforcing too strict controls  on teams can be demoralizing for most, and I’m very opposed to forcing development teams to do things a certain way or dictating how work will be done.  I’m all for ensuring consistency and a good architecture across the application, but this can be done without forcing, controlling, and dictating how it will be done.

To help ensure consistency and buy-in across the board, it is important to consider the following when coming up with your architecture:

  1. How crucial are the recommendations at hand to the business?
  2. Will we see a business benefit by evaluating and following guidelines set around our evaluation?
  3. What is the long term detrimental impact to the software of not doing this or doing it too late in the game?
  4. Will following these recommendations eliminate refactoring costs and technical debt later on?

These considerations will help ensure buy-in and ease collaborative agreements with the team as the team will have a better understanding and vision as to the importance of the architecture to the business.  Having a consistent vision of the business value helps ensure consistency with the architecture moving forward.

Be sure that team members who want to participate can help collaborate on the architecture or guidelines.  This provides ownership by the team members which helps drive and ensure continued consistency.  I’ve said many times that architects should not dictate requirements, rather they should create recommendations facilitated by understanding the software, technology, business, customers, etc.  These recommendations should involve collaboration and review with the other team members before final architecture decisions are agreed upon and finalized.

Let’s look at some real life examples from the field

Example 1) The sales team had trouble in the past selling to some large customers who used primarily Oracle database servers.  The architect discussed the scenario with the business leaders who made up a business case for supporting both Oracle and SqlServer.  Collaboratively with the development team, the architect determined that the code base of the entire application could remain the same, but the data layer could be swapped out to support different platforms.  The business case helped ensure buy-in of an ORM mapping tool to be used as a data layer that supported both platforms.  The team collaborated and evaluated different options finally selecting a tool that met all of the business requirements for performance, scalability, and multiple platforms.  In fact, the tool selected would work easily with SqlServer and Oracle platforms with little overhead.  It was clear to the entire team how important the ongoing use of this ORM framework throughout development would be to the business and that deviating from this could cause considerable damage to the product and business model.  The team was completely onboard with the vision.  In addition, this drove the QA testing team to put controls in their testing processes to ensure compliance with this business requirement.  The QA testing team made sure, as part of their process, to test the software on both SqlServer and Oracle platforms and multiple platform testing environments were created.

Example 2) Another scenario allowed customers to pay for specific application modules , but not others.  It was evident that the architecture needed to adhere to this business model that consistency of the architecture throughout the development process would ensure ongoing compliance with the business objectives.  Collaborative team buy-in of the business benefit of this along with the dependency injection and inversion of control framework ensured that the components being built were modular were able to be swapped in and out of the application easily.   This would also drive QA testing initiatives that would ensure test plans accounted for and tested this modularity as it is a core part of the business model.  From a development standpoint, the team came to a shared vision as to the business reason components are built using this approach.  Teams understand that by not doing this or by deviating from this approach that they are creating a refactoring and technical debt cost to fix this later on.  Of course, teams are free to improve upon this when new functional requirements are added within the iterative development process.

In these two examples, it was clear to see that team buy in and a shared vision were established because the architectural approaches were well thought out, evaluated, aligned to the business requirements, and had a huge cost to change if we got it wrong.  Importantly, the entire team had an opportunity to be involved collaboratively in coming up with the architecture which further strengthened their commitment as they implicitly took ownership among themselves in coming up with the architecture.

It is much more difficult to ensure consistency across development teams without a shared vision about the value being delivered.  For example, dictating certain design patterns to be used over other patterns is a subjective decision that will likely fail buy-in as the business value isn’t clear.  A forced buy-in approach which will likely fail and will likely lead to team demoralization.

Guidance, recommended patterns, approaches, and coding standards can all be put in place.  In reality, it means very little unless we are leading by example and have shown through practice, team buy-in, and business value why we are using said approaches and what the business advantage is.  Instead of working on aligning architecture with business requirements, I’ve seen teams spend weeks coming up with coding practices (how to declare variables, which variable naming pattern to use, etc) for new projects.  The problem is that most people don’t read or care to look at the documents created from these team sessions, and ensuring compliance for compliance-sake can be difficult and is a waste of time.  Even if there is a little bit of value in ensuring how code is written and that variable naming is consistent across the board, I don’t believe documents standardizing the approach provide the value or incentives to do this.

Sometimes, to rescue a failing project, you may need to assert more control and constraints in order to get to a point where the software is coherent and beginning to meet the business objectives.  This is a state that we are trying to avoid by doing our up-front architecture work and ensuring consistency with a shared vision.  However, as consultants, sometimes we are brought into the project too late in the game.  If this happens, trying to fix the solution may require short term measures and controls, but don’t lose sight of the fact that the real value is in ensuring consistent business value through architecture and a shared vision.  There may be a lot of refactoring that needs to be done, but the teams still need to share a vision as to the business value of what is trying to be accomplished.  My experience is that dictating control will only work in emergency scenarios in the short term just to get to a stabilization point, but for the long term, the team needs to work with a shared vision and understanding of the business value to make consistent progress.

Getting the team’s buy-in and creating a shared vision may be a bit challenging and may take longer in failing projects where the vision wasn’t there from the beginning, but it’s the best bet for the long term success of your software and for the consistency and sake of your business objectives.  Once the team has a shared vision and is consistently contributing to the business value, less controls will be required.  Your software will be continuously aligned with your business objectives as both the development and testing teams work together to ensure that your software is adhering to your critical business objectives.

Dan Douglas is a professional independent Software Consultant and an experienced and proven subject matter expert, decision maker, and leader in the area of Software Architecture. His professional experience represents over 12 years of architecting and developing highly successful large scale solutions.  Dan has been the architect lead on over 15 development projects.

Aligning Your Software Architecture With Your Key Business Objectives and Why Your Business Needs It

Abstract

Software development projects need to be aligned with your key business objectives.  The key business objectives that have the biggest cost to change need to be baked into your core architecture as early in the development process as possible.  Key business objectives relating to what the customer is paying for, such as, modularity of components (ex: paying for some components, but not others), performance and scalability, and data accessibility, are a few of a plethora of possible key business objectives that need to be baked into the core-architecture.  Failure to do so can double or triple your development costs, leading to months of refactoring and potential customer and revenue loss.  In agile environments, this article places importance on ensuring that significant core-architecture decisions are not made too late in the game as the myth of “no upfront architecture in agile” is debunked with real world examples where user stories, while aligned with functional requirements, were not aligned with the key business objectives nor the core-architecture.


Aligning Your Software Architecture With Your Key Business Objectives and Why Your Business Needs It

Software projects need an architecture – a core architecture, but defining that architecture and ensuring it meets the business and technology objectives is a bit more thought out than just doing “architecture”.  This article will focus on specific areas in creating a core-architecture as related to key business objectives.

Listing some design patterns and layers for your new system and presenting to the team might have some technical merit, but it isn’t enough if you want to ensure a well thought out and successful software system that meets your business objectives. Architecture isn’t just design patterns, layers, and code design. In fact that’s a very small part of it, if it even qualifies at all. Architecture is more about making very significant decisions that will help ensure the alignment of your completed software with your business and technology objectives.

Creating the right architecture requires business and domain knowledge, product and customer knowledge, research, communication, technological evaluations, technical agility, and expert experience in software development.  The decisions made here will shape the final software solution, so your core-architecture really represents these decisions that have been made during this architecture creation phase.

Now, separate this from day to day software development.  Teams make “architectural-like” decisions all the time when they determine how to implement specific functionality.  Ideas will get tossed around about how many layers of abstraction will be implemented, which design patterns to use, and so forth.  Although some would categorize this solely as design, in a general sense this is still architecture, but maybe not your core-architecture. You could certainly say that not all of these design decisions will have a significant impact on the software.

Grady Booch states that “All architecture is design, but not all design is architecture”.  He also states that “Architecture represents the significant design decisions that shape a system, where significant is measured by cost of change”

Your core-architecture needs to represent those significant decisions and most importantly of all, they need to be aligned with the business requirements.

Some examples of these significant decisions that become part of your core architecture include technology selection (development languages, frameworks, server platforms, etc), application deployment considerations, and technology considerations for key business objectives.

Some examples of architectural decisions to support technology considerations for key business objectives are as follows:

  • The system must work on both Oracle and SQL Server databases. -> An ORM tool is selected that will allow the application to easily swap DBMS vendors.  The ORM tool that has been selected has been evaluated against other tools and also meets the performance requirements. Using this tool will also lead to faster data layer development time over the alternative tools.
  • Customers can use and pay for a variety of modules that need to be plugged in at run time.  The modules need to work together to share information, but also operate independently if related modules are not available. -> Dependency Injection and Inversion of Control frameworks are evaluated against writing an internal DI/IoC framework, and an approach is selected to control how and when the components get loaded and used.  The components will adhere to a specific interface to ensure compatibility and modularity within the system.
  • The system has to be fast – a single implementation must handle a minimum of 100 customers and 5,000 simultaneous users with no performance degradation.  This performance must be maintained with over 100,000,000 database records in the core database tables, and will be measured by… etc… -> Performance considerations relating to how data is retrieved and stored, caching, scalability, and data load operations are reviewed.  Frameworks and architecture decisions related to this are reviewed and selected to ensure that performance considerations are baked into the core architecture. Coding standards and design patterns are put in place to ensure the UI is always responsive and UI data is loaded asynchronously to not freeze the user experience for any amount of time.
  • Customers are paying to be able to access their data in a standard way using 3rd party tools, and want to write automated scripts to retrieve and access this data. -> Business layer will be accessible to and have a corresponding REST API where all customer data will be accessible through this secure REST API.
  • We have licensed 3rd party vendors that pay a yearly fee to write and sell reports to our customers.  We need an open reporting tool with easy access to report data. -> A technology evaluation is completed and multiple technology vendors are evaluated.  The choice is made to use SQL Server Reporting Services to allow other vendors to easily create and sell reports to our customers. Reporting will not be done on the live database to minimize performance impact and mitigate the risk of rogue reports making their way into the reporting module.  A separate star-schema analytics server will be deployed that contains aggregated customer data that is suitable for very fast customer reporting with no impact to the live production system.

As part of this process you may need to include specific details about how this will be implemented, measured, how the testing team will test the performance, the specific technology in question and how it will be used, etc.  Your project will also need a shared vision of this architecture as the success of the project will depend on ensuring this architecture is maintained throughout the development process.

It should be clear to see that the core-architecture will represent these decisions that will have a big impact on the completed software.  Trying to change or implement these core-architecture decisions mid-way through the development process can have vigorous consequences and there will be a refactoring cost that could require months of development time. The worst case scenario is that it could be deemed to be virtually impossible without a major re-write if you have moved too far in another direction.

Could you imagine discovering close to a customer release that your software that was supposed to scale, doesn’t scale?  Or that specific functionality doesn’t work when more than a few users are accessing it at the same time?  It happens often when the software team was not mature enough to ensure that there was a core-architecture aligned with the key business objectives and ensure that the developed software code was aligned with the core-architecture.  These things must be accounted for and mitigated against by your core-architecture and the concepts of the architecture need to be baked into all development decisions made by the team throughout the development process.

If the problems mentioned above happen on your development project it is likely that your core-architecture was incorrect or never established, never followed, or effectively established too late in the development process. These problems alone could easily double or triple your total development costs, cause considerable delay to product shipping, and cost your organization customer and profit losses.

“What about Agile?  We don’t do up front architecture in Agile, we do it as needed during our iterations and we follow the last responsible moment design principle in doing so.”

I’m a proponent of agile methodologies, and I’ve seen the great benefits that effective agile teams can have on a development project, but one thing I’ve seen over and over again in many agile environments is the assumption that less (sometimes zero) time should be spent on the architecture up front. The principle itself is sound as removing most up-front design for functional requirements to be developed as slices of functionality throughout the iterations provides a benefit of not over complicating and over architecting code. However, the assumption that we throw away all up-front design is incorrect. Even on agile projects there are significant decisions relating to your core-architecture and key business requirements that need to be made before development begins and other significant decisions that need to be made as early in the development process as possible. These decisions and related architecture need to be reviewed, updated, and maintained throughout the iterative development process.

Agility is fantastic when dealing with functional requirements and the need to respond quickly to changing functional requirements. However, this agility needs to be kept separate from the up-front core-architectural decisions that need to be made that are aligned with the key business objectives and that will help ensure your software products success and conformance to these key business objectives.

Even in agile environments where YAGNI (You aren’t gonna need it) and “build now, refactor later” are the trends, refactoring code too late in the development process in order meet significant design decisions that align with the business objectives is going to cost you. As mentioned earlier, refactoring costs of months and months of development time is the norm for organizations that didn’t account for core-architecture decisions in their development – especially in agile environments where there is a misconception that these architecture decisions were supposed to be made as late in the game as possible.

In many agile implementations, these core-architecture decisions are made too late as they don’t typically relate to a single user story and they end up having a huge technical debt cost because there is a huge cost to change and cost to refactor existing code once the architecture decisions get made. Days, weeks, or months, can be lost to refactoring.

Another problem on some agile teams is what I call “the race to the finish”. Development teams race to satisfy the requirements of the user stories as quickly as possible to ensure they complete the stories within their iteration and to keep their velocity up (average user story points completed per sprint). And although the functional requirements of the user stories are solid and intact, thought isn’t always given to core-architecture decisions such as modularity, scalability, performance, etc, even though the core-architecture is aligned in parallel with the business objectives. Especially, if the core-architecture hasn’t been or has only loosely been defined, you can expect even less consideration, as the focus turns to completing the work as described instead of ensuring the development is in alignment with the core-architecture. To mitigate this in true agile form, agile teams and the product owners need to ensure that business objectives relating to the core architecture are part of the acceptance criteria for the user stories, and that they are thoroughly tested before the testing team can give the “Ok” on the completed user stories.

Depending on the agile team, how experienced they are, and how senior they are, the focus on how and when design decisions are made can also vary, and not every decision needs to be made during the inception of the project. Typically the biggest decisions with the biggest cost to change should be made as early in the software development process as possible. I would make sure 100% that you don’t lose sight of the “architecture” and the significant design decisions of that architecture and how these decisions are aligning with the key business requirements.

In summary, part of a successful architecture for a successful software product will require significant design decisions to be made up front.  The bigger the cost of change for the design, the sooner the decision needs to be made and implemented within the solution.  Not establishing an architecture can lead to months of wasted development time refactoring code that wasn’t originally aligned with key business objectives such as modularity, performance, and scalability objectives.  In agile environments, it’s especially important to ensure that thought is given to up-front architecture to mitigate the cost of refactoring in the future.  Establishing and maintaining a core-architecture which is tied in parallel with the key business objectives will go a long way to ensuring a successful product rollout.

Dan Douglas is a professional independent Software Consultant and an experienced and proven subject matter expert, decision maker, and leader in the area of Software Architecture. His professional experience represents over 12 years of architecting and developing highly successful large scale solutions.  Dan has been the architect lead on over 15 development projects.

Perhaps you want to learn more about how to establish the architecture, how to present and collaborate with your development team to finalize the architecture and create a shared vision, how to ensure consistency across the architecture, how to bake in non-business requirements such as logging, cross-cutting, and other technical concerns into the architecture.

Or maybe your architecture is established, and your team has done a damn fine job of ensuring the architecture is solid and that it will meet the key business and technology objectives.  How do you maintain this? How do you cope with architecture change when it is warranted or when the business changes?

Stay tuned, as I am writing a series of articles on these topics which will be available in the future.

Creating Enterprise Software Success through Organizational Change

(this new version of this article has been updated as of 11/26/2012 with additional changes and more clarified content since it’s first release on 10/16/2012)

Abstract

Organizations that develop software often succumb to working the way they know how.  This article explores non-technical aspects of software that can be put in place in an organization that will create exponentially better results. Essentially, software organizations need to change how people are working together to ensure long term viability of their software products and their people.  Organizations may need to first flatten their hierarchy a little to start to empower people at all levels.  Following that, organizations need to ensure the architects are working collaboratively with the organization to ensure the developers are focused in the right areas and that the architecture of the product will be sustainable in the long term.  The accountability brought into team members through self-organization and the right leadership will act as a motivator for teams to function at their highest level.  Introducing multi-cross functional teams will further increase communication and knowledge sharing across the organization while creating a platform to easily identify future leaders for the organization’s success planning initiatives.  Once formal and perceived constraints are eliminated and people become more empowered your organization will be better suited to create better quality systems that are executed with high performance.


Creating Enterprise Software Success through Organizational Change

Organizations centered around software often succumb to working “the way we know how” or by following supposedly best practices without considering better practices.  Changing how people are working together, how organizational standards and new initiatives are created and collaborated on, and removing invisible walls can go a long way to greatly improving your effectiveness as a software development organization and essentially create better architected software with better execution while ensuring long term viability.

People tend to work within the constraints of the organization weather the constraints are a formal policy or a perception of how things operate.  Changing how things operate and the beliefs of the people in the organization by empowering them can be a long process, but your organization will be highly rewarded in the long term.

The over importance placed on a hierarchy of people, each responsible, for those below them, can work, but there are many better ways to create a world class software development organization.  Partly because of the hierarchical structure of many organizations, silos and invisible walls are created, and communication isn’t flowing as well as it should be.  Collaboration is usually there, a little bit, but it isn’t as extensive throughout the organization as it could be.  Flattening your structure to an extent that allows for the creation of multi-cross functional teams where people are encouraged to step up, lead, and collaborate on key decisions leads to very effective results and helps the organization to succeed. Naturally, more collaboration will lead to better communication and more innovation throughout the organization which ultimately leads to better results for the organization.

There are many ways to improve enterprise software development and prevent enterprise software development failure.  This article is going to focus on a few people and organizational issues that can be addressed and improved to help create tremendous enterprise software development success. Important things like development process, agile methodologies, testing, pairing, code reviews, enterprise architecture strategy, etc, are all an extremely important part of a successful enterprise software organization, but are out of the scope of this article.

What happens to software with poor communication?

There are too many invisible walls that are preventing communication every day.  People who don’t know each other well, don’t normally communicate very well either.  This is very evident in enterprise software environments where teams of 50+ developers may only know a few of each other very well.  They could be eight feet away in an office, or 2000 miles away in another country, but there is a slight discomfort with developers to voluntarily work together when they don`t know each other well.  Especially, they will be less likely to collaborate, and much less likely to ask for help when working in the same area that the other person is much more familiar with.  Even when they do work together, it isn`t as effective as it should be.  This leads to inferior quality code and product, because the time being spent on development isn’t as productive as it should be and knowledge transfer isn’t happening as it should be.

Limit The Hierarchy

A limited organizational hierarchy approach helps flatten the structure a little bit and open up doors for communication and for more decisions to be made collaboratively.  Structure is still important, and ultimately you need specific people who are each responsible for their specific areas, however flattening it a little bit allows good decisions to be made at all levels.  It helps teams to become more empowered to make their own decisions and to work and collaborate the way they need to. This still keeps some much required structure and focus in place, but promotes good decisions to be made at all levels.  Bottom up and top down. 

The Architects

Consider removing traditional Software Development Managers and Software Development Team leads and replace with well-rounded Software Architects.  The architects are the ones that need to drive execution of the software and are ultimately responsible for the design, roadmap, and value delivered by the software developers.  The architects will have the most exposure to developers and senior management, as it is a requirement to collaborate with executives, product management, development, and other areas in order to properly drive execution.

Ultimately, this level of collaboration by the architects will ultimately benefit the organization by allowing them to come up with the best approach.  This will yield a much better and thought out, future-proof technological decision than if the architect was to come up with the technology requirements individually, which often occurs in ego driven organizations.

You don’t need the title “Software Architect”, but the role should be as described here.  It’s not only a senior developer who understands software architecture, design patterns, etc –  and has architected systems or features in existing applications.  To me that implies a great skillset for a senior developer and a skillset that is sometimes difficult to find.  But to be an architect working with large software projects at the enterprise level, you need a broader range of skills.  The architects also need to be able to motivate and lead a team.  By leading a team, I mean encouraging process improvements upon the team, but leaving the team to decide the organization and completion of the work, letting the team work collaboratively to get things done, but constantly raising the bar of the team so that they continue to excel and so that they don’t fall into a mediocre cycle of maintaining or lessoning their current level of productivity.  The Architect has to be able to lead in such a way that allows the team to understand direction, priorities, and to allow the team to focus on the right objectives.  In the end, if you call this role a “Manager” or an “Architect”, it doesn`t matter – the role of the position is what is important.

Self-Organizing Teams and Accountability

Even within teams there are often invisible walls or silos.  Self-organizing teams go a long way to help ease this and improve software greatness as well.  For a self-organizing team to function at a high level, you need to ensure that people are working together as a team and focusing on winning as a team and accepting losses as a team.

Self-organizing teams work best when they are empowered to work as a team and there isn`t a feeling of an established hierarchy.  There aren`t dedicated team leaders, as leadership will be shown within the team and everyone on the team has a voice and has a chance to step up and lead on decisions.  This greatly improves knowledge sharing as well as the team is working together very closely and knowledge gets shared throughout the team more easily.  Typically a well-functioning self-organizing team operates at a very high level.  They fail as a team and succeed as a team, and therefore work together to do what it takes to improve, succeed, and meet their goals the way they need to while collaborating and helping their fellow team members.  Members of self-organizing teams feel accountable to the rest of the team.

Self-organizing teams are great and highly recommended, but ultimately they still need external motivators to make sure they are continually raising their own bar and to ensure they are focusing on the right things.  Architects along with the rest of the leadership team need to play a role in this.  While the Architects are ultimately responsible for the delivery of the software product from a technical perspective, the accountability for getting the development done comes from within the development teams themselves.  This accountability that comes from, and grows from within, a functioning self-organizing team greatly encourages individual team member productivity and responsibility for getting the work done the best way possible. This is why Architects, in the definition for this role, can’t just be regular senior development guys who understand software architecture.  They have to understand how the aforementioned things, along with execution, are really important if you want a successfully architected piece of enterprise software.

Key Software Decisions

When making key decisions that will affect the software, try to involve a small team to discuss.  Get the biggest experts across your organization, preferably from all business areas impacted, and gather around and discuss ideas.   Having a process dictated by a group of senior level individuals without adequate consensus can completely demoralize a development team and lead to gaps in the process or an ineffective process altogether.  I am not saying to eliminate leadership from making key decisions, because your leadership essentially should be very qualified to make decisions.  However, the best leaders draw on the experience of team members across the organization to make the best decisions.  Enough feedback must be solicited from the organization to ensure that the decision or initiative is the best approach and will have the biggest impact.  Without talking to the team, getting input, a consensus, and feedback, even good leaders will make bad decisions or create useless initiatives and processes.

Certainly, there are sometimes quick decisions that need to be made solely by those that are qualified to make them.  This is ok as well when it is necessary and doesn’t necessarily warrant the overhead in getting others involved, but always make sure that when this happens that the door is open for improvements and ensure that any member of the organization is welcome to help improve the process or help remove the existing barriers which lead to the process to begin with.  This is why we need a culture of open communication.  Typically, when decisions are made this way, they have an impact from the get go, but because they didn’t necessarily have the buy in from the organization, suggestions and alternatives drawing from the experience of others in the organization were not considered which could have an even bigger impact. Don’t form a committee or introduce unnecessary overhead, but just make sure you get the required input you need to make the best decision.

Once new processes or process improvements are in place, the rest of the organization needs to know why they are in place, what they are trying to solve, and know that anyone in the organization is welcome to further discuss this openly with the stakeholders.  Inviting that communication goes far in helping establish trust and helping the people of the organization understand that every process is open for discussion and improvement by anyone.  It removes tension that many people feel when a process was jammed down their throat without any collaborative input from them.

Rewarding Change and Improvements

When someone takes the initiative to change a process or suggest a process improvement, make sure that that individual feels rewarded.  Even saying something like “That’s a great idea and I think it will work well” goes a long way toward building further trust and showing the rest of the organization that this is a company that takes pride in the collaboration, input, and improvements lead by any team member.  Ignoring input of contributors, good input or bad, is the absolute worst thing an organization can do.  It creates tension, inhibits growth, and absolutely ensures that the people trying to contribute will feel unvalued.

Also, consider giving random rewards to team members who are continually contributing and stepping up above and beyond their required responsibilities.  This is important, so that your key people are always rewarded and feel important, and it will help to motivate others as well.

If you’re a thought leader (or want to be) in your organization and someone sends you an email asking for advice or offers an improvement suggestion, talk to him or her.  Or at least send out an email suggesting you meet to discuss in the future.  It shows appreciation and it decreases invisible barriers.  It really doesn`t matter what your position is, if you don’t want people to communicate with you or suggest improvements, change your attitude because you are creating a new invisible wall and it will slowly drag down those people around you.  This goes for any form of communication – don’t create invisible walls – work hard to remove walls and increase communication and collaboration.

The Right Leadership

Architects and your other senior people need strong specific knowledge in the technologies being used and they need to pick up, learn and understand business requirements quickly.  They need to be able to form solid relationships with Product Management, Sales, QA, and Executives to get the big picture and to use that to help create a solid architecture, foundation, and technology roadmap with the rest of the team.

Understanding, learning quickly, team leadership, and strong technical knowledge need to be in place.  So, if you have a software leadership team that just talks, manages, dictates, or doesn’t collaborate, they may be better suited to another role or kept on to help them grow into a role where they are working better with the rest of the team.  However, the point is clear that they aren’t ready and they wouldn’t be effective leaders in the organization.  Notice how I mentioned “just manages”?  I purposely used “manages” as a term relating to managing people.  Aka – making sure “their people” are working, giving reviews related to their performance, checking their attendance, ensuring processes are being followed and all of the things that “managers” do.  The point is that successful product and development teams need leaders and not managers.  Managing doesn’t improve process, but leadership can drive improvements.   Some management may be required by a leader, but leadership should be at least 10:1 the responsibility of management.

Multi Cross Functional Collaboration

I’ve created the expression “Multi cross functional collaboration” as a way to define how to make cross functional teams more collaborative, and even more engaged with other cross functional teams and other parts of the organization.

When people in the organization don’t know each other, it’s a problem.  In fact, it’s a bigger problem than most organizations realize.

In order to remove barriers to communication, start by getting everybody to know each other.  In Agile environments, Agile teams should already be cross functional teams.  You need developers, quality assurance testers, business analysts, product managers, and potentially others, all working together on the same team.  This allows the process to flow smoothly from product inception, development, to testing.  However, even in this environment you tend to still create silos where communication becomes blocked as there are typically barriers of communication between members of this team and members of other teams.  This needs to be addressed, because in software, especially enterprise software, a siloed team still needs open doors to the other teams and to the rest of the organization.

If you expand on the idea of cross functional teams and create the multi-cross functional team, essentially, cross functional teams across your cross functional teams, you are helping to bridge the communication gap by exposing people within existing teams to people in other teams.  You are bringing together people who are experts in their specific areas to work together and come up with the best solutions.  You are having them work together more closely, sharing knowledge, collaborating, and building new inter-enterprise relationships. You still keep the original teams intact, but you create other permanent or temporary teams that work together to help improve process or solve different problems.

Multi Cross Functional Teams In The Real World

As an example, you could have five software teams that are cross functional – three developers per team, one QA tester, and a business analyst.  This is great, because the teams involved in delivering pieces of the software product from end to end are working together as a team.  But often the teams still aren’t working with the rest of the organization that well, and this needs to be addressed.

Think about this real world example.  You want to implement a new extension to your product and want to figure out the best approach or brainstorm ideas about it. You don’t want to develop a solution yet, but you just want to gather ideas and determine architecturally what is the best approach to proceed, what technologies could be used, and how to improve existing standards to better facilitate future development.  This new potential product extension will reach across several software disciplines in the organization and across the functional areas of many existing cross functional teams.  What do you do?  One approach would be to ask the architects or leadership how to proceed, but a better approach is to create a team comprising of individuals across different existing cross functional teams at all levels (not just leadership).

To create the new team, invite a few key people who definitely will be able to provide big value to the team.  Encourage others to volunteer to step up and join.  Now you’ve got a new cross functional team that spans multiple existing cross functional teams.  And you’ve got a handful of people most likely who are now working together  who haven’t likely communicated much in the past due to existing barriers to communication and also because they just don’t know each other that well.  The team can work together to come up with solutions to the problem they are trying to solve and also help share their progress with the rest of the organization.  It also gives people who wouldn’t normally have a chance to step up and be a leader to step up and help collaborate and lead direction on another team.  This is a huge step for an organization for relieving communication problems over the long term while getting new and important side projects started and finished, and exposing team members to new people in the organization (internal or remotely).

Organizational Growth Through Engagement and Empowerment

Think about how engaging and empowering individuals to step up and contribute in more areas of the organization will allow your people to grow, and in turn allow your organization to mature.  It will help prevent boredom or developer burnout by exposing people to new people, ideas, technology, and teams.  Now, think about this.  This is a tremendously huge opportunity for the organization to fuel their succession planning.  Now you can see who is working on different teams, who is stepping up, who is taking on a leadership role for certain initiatives – even where it wasn’t even a requirement for their current role.  This is huge for an organization trying to figure out who their best people are and help them figure out where they can fit them in to the succession planning process. You don’t want your best people to leave to pursue greater opportunities in other companies, so empowering people with a chance to step up, collaborate, and lead, will leave them with an avenue to be recognized in the organization as a leader or potential leader, collaborator, and asset to the organization.

This also helps go towards the problem of developer boredom which plagues software development projects. If you think developer boredom is caused by bad developers, think again.  Boredom creates mediocrity and inevitably even extremely talented people become less productive.  Unexciting or repetitive work typically creates the kind of boredom that can usually be relieved by leaving the organization and starting fresh on a new project.  Get my point?  Elevating boredom is huge for a company, and even allowing your people to spend part of their days exploring new ideas and projects (maybe as part of another team) that are not directly related to what they have been doing day in and day out for six months is a huge step for morale and will make their regular work much more productive as a result.

Other examples where creating a new multi-cross functional team is beneficial to the organization include:  Nailing down some pairing and communication processes, R&D on potential new technology to incorporate into the project, exploratory research projects, defining new architectural standards or approaches, new development initiatives such as unit testing or code reviews, etc.

The point is to get the ideas started and then make sure that everyone in the organization knows what is coming up in terms of new initiatives or teams.

The Best Part – Employee Retention & Succession Planning

Great developers work hard, great developers get excited about new technology, product creation, and innovation.   Great developers generally have a spirit of entrepreneurship towards their work in the sense that they love to try new things, fail, learn new technologies and try them out.  This excites the majority of developers and ignites a passion in them.  By including them in different projects, and giving them the opportunity to step up, they also gain a sense of importance to the organization.  Think about how great this lends itself to employee retention.

Exposure to new technologies, people, and helping define processes shakes the work day up which helps prevent the mundane problem of boredom when working on the same things for too long.

Remember the fantastic side effect?  Succession planning.  This helps create a roadmap for succession planning in your organization.  Now, multi-cross functional team member’s have an even bigger opportunity to shine by stepping up and joining (or better yet, starting) other new cross functional teams.  In these teams they can prove their ability to work in different facets of the company and communicate, lead, and collaborate with different teams and people all while increasing their own technical, and communication skills.  This helps management decide who should, let’s say, become the next architect, or lead a new software development initiative while also raising the skill level of everybody else who has been involved.  It puts sustainable succession planning into motion within your organization.

Finally

Ultimately, the role of leadership is to create a vision and have buy in with the organization so everyone has the same vision, recognize the contributions of the rest of the organization, ensuring that people are collaborating to the best extent possible, eliminate invisible walls and silos, all while sustaining long term growth and evolution of the organization.  The accountability brought into team members through self-organization and the right leadership will act as a motivator for teams to function at their highest level.

Using mutli-cross functional teams, and encouraging decisions and improvements to be made at all levels while getting the input of the people across the organization before implementing big initiatives will be valuable in ensuring that the best and most thought out decisions can be made.  This helps empower people and leads to much better succession planning.

The right people are key, so having very smart people at all levels is a requirement to correctly implement the ideas presented in this article. It`s hard to build a relationship of trust and empower people within your organization if you do not have the right people who are intelligent, will step up, contribute great ideas, and open up doors of communication. Eliminating the formal or perceived constraints that people are working within in the organization can be a long process, but once you empower people your organization will be better suited to create better quality systems that are well executed.    The advice and topics discussed in this article can go a long way to assist in achieving these goals and to help create an even more successful software organization.

Dan Douglas is a professional independent Software Consultant and an experienced and proven subject matter expert, decision maker, and leader in the area of Software Development and Architecture. His professional experience represents over 12 years of architecting and developing highly successful large scale solutions. Dan also believes that properly empowering teams with trust and responsibility yields the greatest results, and that creating better software goes well beyond just writing better code.

The opinions expressed in this article are my own and are based what I have done, what I have seen, and what I have learned while leading and working in many different team environments.