In a recent DevOps.com webinar, Anders Wallgren (Electric Cloud CTO) and John Willis (Electric Cloud advisor and co-author of The DevOps Handbook) discussed the “7 Deadly Diseases of DevOps.” In the first post of the series covering each of these diseases, we shared their expert insights on invisible work and how it hinders IT organizational success. Willis mentions the fact that most large organizations only have visibility into 50% of the work that is being done, which in turn diminishes software delivery efficiency and quality. When half of the work is being done in the dark it becomes close to impossible to know what is working well, what isn’t and how best to improve. Willis and Wallgren cover more warning signs and remedies of invisible work in the first post of this series.
Let’s quickly review each of the seven DevOps “diseases” laid out by Willis and Wallgren and dive into the next topic, management system toil:
2) Management System Toil
3) Tribal Knowledge
4) Misalignment of Incentives
5) Incongruent Organizational Design
6) Managing Complexity
7) Security and Compliance Theater
The second disease, “management system toil” refers to the numerous, disparate management systems (JIRA, SharePoint, Remedy, etc.) that organizations employ across teams and functions, creating behind-the-scenes chaos and bottlenecks. Here’s what Wallgren and Willis had to say about this disease and how it is negatively impacting IT organizations today:
Deadly DevOps Disease #2: Management System Toil
Willis defines the ticketing strangler pattern as a method for solving management system toil: “A ticketing strangler pattern is where what you start taking all these disparate systems and drive them towards one system. Hopefully, you’re driving them towards something like a development methodology system, maybe JIRA or whatever your weapon of choice is there.”
Referring to ticketing systems, Wallgren says there isn’t much value: “To me, many ticketing systems are, and when I say ticketing systems, I mean more service management systems, just really expensive Post-It notes. Ticketing systems are just another opportunity for miscommunication, another opportunity for delay.”
Really, all of these diseases are intertwined and funnel down into each consecutive one, says Willis: “What you’re seeing here are these kinds of diseases funnel their way down into uncovering knowledge about where you are as an organization and then actually help start creating a true dotted line, hopefully eventually a solid line, to getting to the kinds of principles and practices that are well defined in our community as high-performance.”
Wallgren explains how using a self-service ticketing system can help remedy management system toil: “If I need an environment set up, I don’t file a ticket to ask for it and then wait until somebody gets around to doing it. I go to the self-service system and pick from a pre-approved menu of environments with the authentication and authority to do so. Then, I add in my preferences and five minutes later I have my environment. In the meantime, in the background, the ticketing system has been updated. So, we have an audit capture of who requested it, when it happened, etc. It’s not so simple as filing a ticket and then it happens. It’s asking for it to be done, it happens automatically, and the audit action happens automatically.”
Willis ties in how invisible work effects monitoring of management systems: “How do you do monitoring? Who do you send alerts to when you have monitoring? I’ve talked to people that run service catalogs or configuration management databases (CMDB) and I’ll ask how accurate they are. A lot of people say their CMDB’s are 30-40% accurate. The reason they’re only 30% or 40% accurate is because you’re only capturing about 50% of the work.”
For more insights into the seven deadly diseases of DevOps, watch the full webinar.
Stay tuned for our next post on Tribal Knowledge!