Archive for July, 2016

IMG_0077

Agile values and principles are the core foundation by which Agile organizations operate and make decisions. Everything we do is based in these.  With that being said, viewing every principle through a holistic perspective is absolutely necessary.  Every word in the principles we live by has value and impact.  So, when we reduce a principle to a three word summary, I believe we do ourselves a disservice.  This practice often results in focusing on part of the principle without the balance of the other side.  Through this oversight, we inadvertently create environments where there is unbalance that leaves people frustrated and confused.  They begin to believe that Agile is the problem.  But, the real problem is our failure completely embrace the Agile values and principles and settle for anti-patterns instead.

Today, I’d like to take a deeper look into Agile Principle #8 which states:  Agile processes promote sustainable development.  The sponsors, developers, and users should be able to maintain a constant pace indefinitely. 

People often refer to this principle as the sustainable pace principle.  The most common description given of how we practice this principle is that the development team should not be expected to do more work than they can complete in a normal business day.  We don’t want people working 70 hours a week because they are forced to do more work than is possible during a normal work week.  Working at that pace is something a team may be able to do for a sprint or two but they cannot work at that pace indefinitely.  When people are tired and overworked they make more mistakes and it actually slows down their ability to produce work.  It also impacts motivation.  When people are overworked and have no work/life balance motivation dwindles.

But there’s another part to this principle that I don’t hear quoted as often.  It’s the part that talks about the constant pace at which the sponsors, developers, and users should be able to maintain.  This is about the consistency of our delivery, sometimes referred to as predictability.  Developers should be able to trust that sponsors and users will allow them to work at a sustainable pace.  In return, sponsors and users should be able to trust that developers will consistently provide a continuous stream of valuable software.  The team has a responsibility to be transparent with the sponsors and users regarding how much work they can complete in a certain timeframe.  They also have a responsibility to be transparent when the forecast must be changed along the way due to new information or unforeseen problems.  This gives the sponsors and users the ability to communicate and make decisions regarding the impact of the forecast change.

How does this impact the way the team conducts planning and communicates their forecast?  Teams should plan for as much as they can realistically complete and communicate that forecast.  Then, they should strive to complete 100% of their forecast every sprint.  If something happens to prevent the completion of the forecast they should communicate as soon as feasible to stakeholders so they know what to expect.

Should teams forecast 125% of what they believe they can realistically complete and be happy if 80% of the work gets finished?  No.  Why?  First, because it sets unrealistic stakeholder expectations to communicate more work than the team can realistically expect to finish.  Second, because it contributes to a lack of trust between the stakeholder and the team when the team keeps promising work they consistently don’t deliver.  Third, because the extra time planning and tasking stories that aren’t likely to be worked creates waste and adds unnecessary time to the planning process.

Then what do we do with “stretch” stories?  It is my belief that “stretch” stories are not a part of the forecast.  Plan and communicate what you believe you can complete.  If the backlog is groomed properly it will always have at 1-2 sprints worth of work in “ready” state.  So, if the team runs out of work they can always agree to pull in another story.  The solutioning and tasking for that story can take place when the decision to pull it in happens.

If the team consistently gets 100% for 3-5 sprints, stretch yourself and bring a few more points into your sprint forecast.  It may take you a couple of sprints to get to 100% again but it will stretch your ability to produce work and push you to incorporate practices like automation in order to move faster.

There should be an understanding that no team will always complete 100% of the work forecasted.  This is another part of the concept of trust and transparency.  Stakeholders and customers trust that developers will always strive to complete 100% of the forecasted work.  Developers trust that when something happens and they can’t deliver 100% and communicate openly to stakeholders and users there will be grace and understanding extended.

 

IMG_3265

One of my favorite things about being and Agile coach is connecting with the Agile community through conferences, meet-ups, and other networks.  Because of these connections I get to interact with  Agilists all over the world.  Over the past few months I’ve noticed a concerning trend coming from the Scrum Master community.  They are telling me with excitement, “I’ve finally worked myself up to two teams!”  Some have said they are now working with three or four teams.  The thing that concerns me is that they seem to view spreading themselves across multiple teams as an accomplishment.  I am hearing pride in “being busy” and “being able to handle more” and that tells me that we still have work to do.  It tells me that there may still be an anti-pattern running rampant in our Agile organizations telling us lies.

The belief that “the more I can handle and the busier I am the more valuable I am to the company” is left over from days when sustainable pace wasn’t a part of the culture.  The truth is that busy does not equal productive.  Our highest priority is to satisfy the customer through early and continuous delivery of valuable software.  It is not to ensure that everyone is at least 100% (or more) utilized.  Agile processes are supposed to promote sustainable development.  The sponsors, developers, and users should be able to maintain a constant pace indefinitely.  When we are running at 100% (or more) of our capacity we cannot maintain that pace indefinitely.  At some point, we burn out mentally, physically, and emotionally.  We cannot afford choose utilization over productivity.  Our primary measure of progress should be working software, not how much more we can get done with fewer people.  The efficiencies in an Agile organization don’t come from piling more work on fewer people.  They come from improving our technical practices, increasing automation, increasing quality, lowering technical debt, collaborating, and learning to continuously improve our processes.  These things give us the ability to produce more without adding employees because we stop tripping over ourselves and can run along a clear path.

I read the following question from a user on stackoverflow.com:

“Does running your servers at 100% CPU usage cause any issues or is it just good CPU utilization?  My servers have 8 physical cores constantly running at near 100% for “open hours”/10 hours per day.  The program is architected to run on 8 threads – and it fully uses them. Performance is good but the infrastructure guys are worrying about the “maxed out servers.”  I think it’s just good use of available resources. What’s the point of having lots of core if they are not all fully utilized?”

The problem with this line of thinking is that when resources are fully utilized they don’t get more done.  Contrarily, less gets done.  They move slower, wait time increases, and so do errors.  Here’s the response someone gave to this question:

“Almost without exception it causes issues, or will cause issues down the road (as demand grows).  100% CPU utilization on a web service server is not good.  If your CPU utilization is at 100% it means that each time the server gets a new request there is a 100% chance that the work will have to wait some amount of time before the server gets started on it.  The typical performance sweet spot is about 70%.  Does that sound low?  If so, remember that 70% utilization doesn’t mean that 30% of the CPU is being wasted.  Instead, it means that 70% of the CPU’s capacity was used over a sample period.  For CPU measurement metrics, a sample period is something like 2 seconds.  During that 2 seconds the breakdown of that 70% is uneven.  In other words, it may be something like 100% for 1 second and 40% in one second.  For short bursts like that, 100% utilization is okay because we know that if a piece of work is delayed it is only for a brief period. (One that won’t make the human waiting upset.)”

I’m wondering, if we adhere to this rule with our hardware resources, why don’t we realize that the same rule applies to our human resources?

I’ve been in the position where I was a scrum master on one team doing an excellent job.  I knew the pulse of my team and they were growing rapidly and performing better than ever.  Then, I was given a second team.  Sure, I had enough down time in my average week to handle facilitating Scrum events for two teams (in theory) but because I was toggling between two team rooms I missed a lot on both. On sprint end/start days I felt very pressured.  I ran from one retro to the next on and often couldn’t compile the improvement plan into a consumable format until two days later.  I fell behind updating information radiators and had less time to think analytically through what was happening with each team.  Over time I saw that both teams were maintaining, even growing some, but the rate of growth was slower than when I had only one team.

Then, something tragic happened.  I was “doing such a great job” that I was asked to take on two more teams for a month to fill a hiring gap.  I felt like a total failure.  I had to choose which teams I was going to work with and leave the others stranded.  I had no clue what was going on in any of the four teams because I wasn’t spending enough time with any of them to catch the important conversations.  My teams all felt abandoned by me and had to pick up the slack felt by my absence.  While in my manager’s eyes nothing fell to the ground because my teams were mature enough to fill in the gaps without me, my teams felt all the pain and none of the benefit.

I learned a very powerful lesson through that experience.  Being utilized at 100% (or more) capacity didn’t make me a super Scrum Master.  It made me a terrible Scrum Master.  On a ledger somewhere it may have looked like the company saved money by utilizing me to full capacity but the impact of the hidden cost was much greater than the financial gain.  We would have done better to allow the third and fourth teams to work without a scrum master for that month.  Instead, we caused four teams to operate without a scrum master by spreading me too thin.

What message do we send as an organization when we tell our teams we expect them to plan their sprints at 125% of their capacity?  It’s a message that says we do not value sustainable pace.  What message do we send when we tell our employees we want them on multiple teams so we can fully utilize their capacity?  It’s a message that says we do not value sustainable pace.  What message do I hear when Scrum Masters tell me proudly that they are working on multiple teams?  I hear that they have forsaken the Agile principle of sustainable pace.  I hear an anti-pattern.  It makes me know that though we have come far we still have more work to do before becoming truly Agile.