SmartDraw recently commissioned a survey on the topic of Visibility for DevOps. The first few posts discussed the overall findings and dove into a few interesting findings. In this post, we'll explore the challenges DevOps professionals face when trying to provide real-time DevOps visibility.
Our survey split out those organizations that are doing the very best at providing DevOps visibility. What's interesting is that these DevOps professionals, while doing well, still find the process challenging. Here are the top challenges they face:
Reporting Takes Skilled Workers Away from More Important Work
This makes sense. DevOps teams are - by definition - small, tight groups. Jeff Bezos famously insisted that teams should be small enough to feed with two pizzas. His logic? Larger groups are lousy at communicating and that slows everything down. This is part and parcel of the DevOps philosophy and explains why DevOps teams tend to be small and agile.
The flip side of small teams, however, is that when you task a team member with something like producing real-time DevOps visibility, you're creating a log jam. Important work gets put on hold while this team member does the visibility work.
Reports are Difficult to Read and Understand
In theory, this is potentially true for any team that creates reports for any type of projects. But it is especially true for DevOps because so much of what DevOps does involves complex information based on hierarchies, relationships and flows. Simple spreadsheets or tables are poor substitutes for the purpose-built diagrams that have evolved for these disciplines, but they simply aren't diagrams that a typical worker can create without purpose-built tools. So, while a table or spreadsheet is a clearly inferior way to show certain types of information, it is often the only practical tool.
Reports are Based on Stale Data
This is clearly a problem that any team would want to avoid. But with DevOps it is an especially damaging issue. DevOps operates at the extreme limit of how fast humans can work, and thus the information about projects is inherently transitory. For example, year-to-date sales information that is two days old is quite acceptable, yet two-day old application delivery status in an organization that deploys 3 times a day would be hopelessly out of date.
The challenge, however, is that DevOps information is pulled from multiple disparate applications. It is too difficult and time-consuming to pull new, fresh data every single time a new report is needed. Thus, the reality is that DevOps reports are often based on stale data.
The bottom line is that while some organizations are able to create DevOps visibility, it isn't without a lot of effort, and comes with some serious side effects.