Key Takeaways from Continuous Discussions (#c9d9) Episode 74 –  DevOps for Mainframes

In a recent Continuous Discussions (#c9d9) video podcast , expert panelists discussed DevOps for Mainframes .
Our expert panel included:Robert Stroud , principal analyst at Forrester; Glenn Everitt , technical strategist at Compuware; Santosh Vaswani , technical manager of validation & verification at eInfochips; and, our very own Anders Wallgren and Sam Fell .
During the episode, the panelists discussed whether you can and should modernize your mainframe with DevOps, DevOps challenges presented by mainframes and key patterns. Continue reading for their insights!

Should you DevOps your Mainframe?

When we first started talking to people about #DevOps for #mainframes we got a lot of weird looks 😂 @glenneveritt #c9d9 pic.twitter.com/hOn5qtOuL6

— CloudBees (@electriccloud) July 25, 2017

Almost by definition, anything running on the #mainframe is mission-critical, per @anders_wallgren #c9d9 #DevOps #business pic.twitter.com/rk4YYyuk5y

— CloudBees (@electriccloud) July 25, 2017

Stroud is in full support of DevOps for the mainframe: “The mainframe is everywhere and it absolutely has a part to play. We've seen DevOps processes extend to the mainframe allowing development teams to get quite agile and deliver velocity. They're actually learning it by doing smaller consistent changes and they’re doing it better.”

DevOps on the mainframe makes a lot of sense according to Everitt : “A lot of people complain about the cost of a mainframe. I keep finding out is it's not so much the cost of the mainframe as it is the speed of the mainframe. That is one of the main reasons you want to do DevOps on the mainframe in order to get speed in your processes and get to production faster.”

Vaswani is in agreement with Stroud and Everitt: “If you look into the automation side, there are very limited toolsets available for the mainframe. We have to come up with the tools and more tool chains in which we can completely implement the DevOps process for the whole mainframe as well as mainframe applications.”

At its core, the mainframe is software, says Wallgren : “In some cases, it is legacy code produced decades ago. It's a unique place culturally and technically. That said, this is just software, whether it runs on a phone or a mainframe or a drone or whatever. So, we treat that the same way we would any other decade-old legacy code.”

DevOps for Mainframe Key Challenges

Some things are on the #mainframe and some things are migrated to other environments - another challenge of #DevOps per @ei_santosh #c9d9 pic.twitter.com/xvWPAIpAsW

— CloudBees (@electriccloud) July 25, 2017

"If it's not automated it's not going to happen consistently, on time or correctly" @glenneveritt #c9d9 #DevOps #mainframe #automation pic.twitter.com/04awcEq5Lf

— CloudBees (@electriccloud) July 25, 2017

Coordinating pipelines with different releases is challenging, per @samueldfell discussing #DevOps for #mainframe on #c9d9 pic.twitter.com/FOcsbVZMiu

— CloudBees (@electriccloud) July 25, 2017

Everitt on the importance of doing DevOps in the mainframe: “If you have a distributed set of systems that are all trying to do DevOps and you're not doing DevOps in the mainframe that means you're not going to be able to deploy with everything else you deploy. You have to get it deployed in the first place if you want to test it.”

Innovation at the mainframe level is not really happening, states Wallgren “We don't necessarily have a line of people at the door wanting to learn these things. I'm not 100 sure, but the amount of innovation that's happening inside of the mainframe community is probably not that great.”

Vaswani notes the challenges of running multiple environments: “There are the different product pipelines running at the same time. Some of the things are on the mainframe and some of the things are migrated to some different newer environment.”

Fell highlighted deployment challenges: “You have multiple different types of applications that are dependent on the mainframe. How do you coordinate all of those different pipelines with all of those different releases? Even when there may not be anything major being changed, you have to still do integration testing and make sure that those things are going to work together.”

Stroud offers a counter view, suggesting that some challenges are false: “The challenge I hear is that code is being built with business logic and everything is all integrated into one in terms of old code. These are bad coding practices we gave up 20 years ago. Those challenges can be overcome. We're now well into the future - hardware's reliable now, software is reliable. We've got good tools and the challenges we are stuck with are a bunch of myths from the past.”

Patterns for DevOps for Mainframe

We've got to attract #developers to the #mainframe environment and apply modern techniques #agile #devops @RobertEStroud #c9d9 pic.twitter.com/U7UT1LiiSu

— CloudBees (@electriccloud) July 25, 2017

There's the "millenial underground" who want to learn #mainframe on their own per @glenneveritt | #c9d9 #DevOps pic.twitter.com/TNCP4Azri9

— CloudBees (@electriccloud) July 25, 2017

Reducing release time is one of the ROI's @ei_santosh measures in their #DevOps transformation | #c9d9 #mainframe pic.twitter.com/psFdbQ6W4f

— CloudBees (@electriccloud) July 25, 2017

Maybe try giving your team(s) an assessment similar to this assessment Vaswani uses: “What we do first is a DevOps assessment. The DevOps assessment gives us the status of where we are standing currently - are we on the level one, two, three, four or five? We call them initial, managed, defined, measured and optimized.”

Fell mentions that having subject matter experts (SME’s) on mainframes could be helpful: “One of the other patterns I see with a lot of our customers is they've created a group of SME's around DevOps. I've seen some companies where they created a separate team that's responsible for deployment and testing and that usually seems like that creates another silo which is bad. But this center of excellence team that goes from group to group and helps them out and on that DevOps journey, which seems to be a model that works well.”

It all comes down to consistency, says Everitt : “You got to be able to build it consistently before you can deploy consistently, and you got to be able to deploy it consistently before you can test it consistently. Going down that path frees up resources to help you do that next step.”

Practicing DevOps can lead to better code, per Stroud : “When companies leverage DevOps in the mainframe and they start to look at their code and the way they transition it, we clearly see a change in practice. This change in practice reflects in better code with greater quality.”

Isolate your mainframes, advises Wallgren : “Provide a little bit of isolation between the core mainframe implementation and its clients by using service facades or API's. This will give you a little bit of isolation where maybe you didn't have it before. Then you can start to rearrange things under the covers in the mainframe.”

Watch the full episode:

Want more Continuous Discussions (#c9d9)?
We hold our #c9d9 podcast every other Tuesday at 10 a.m. PST. Each episode features expert panelists talking about DevOps, Continuous Delivery, Agile and more.
Check out all past episodes and panelists here .