DevOps.com editor-in-chief Alan Shimel sat down to speak with Dr. CJ Paul, Distinguished Engineer, DevOps Methods & Tools, IBM Cloud. CJ is someone we have interviewed previously on video at IBM InterConnect and is always a great source of information. In this interview Dr. Paul discussed DevOps, ITSM, Cloud, Bluemix, Bluemix Garage method and more.
As usual, the streaming audio of our discussion is below. You can also find it on SoundCloud and iTunes. Below the audio player is the transcript of our conversation so you can follow along. Enjoy!
Shimel: Hi, everyone. Alan Shimel here for another DevOps Chat on DevOps.com. Thanks for joining us. I’m happy to be joined on this episode of DevOps Chat by a friend of mine CJ Paul of IBM. CJ, are you there?
Paul: Yes. Hi, Alan. How are you?
Shimel: Good. Good to have you on DevOps Chat with me today, CJ. Of course you and I have done several video interviews from various conferences we’ve both attended. But you know we haven’t had the pleasure of having you here, so welcome. CJ, I’m gonna bet not everyone in our audience knows who you are. Why don’t you give a quick little background on who you are, what you do, where you work and so forth?
Paul: Yeah. Definitely. So my name is CJ Paul. I work at IBM. I’m a Distinguished Engineer in the cloud unit. And focused on DevOps methods and tools. So some of the things that my team puts out are the Bluemix Garage Method. It’s out as a website at IBM.com/devops/method. Then we also work on the Bluemix DevOps services and you know I work with a lot of clients as they do their transformation to the cloud. And trying to develop DevOps practices along the way.
Shimel: Great. So CJ, you know just kind of off the beaten path here, but we’ve had several guests on DevOps Chats that are distinguished engineers at IBM. And for our listeners who you know maybe are not familiar with what that means and it’s more than just a pretty title. Can you give our listeners a little bit of insight into what it is to be a distinguished engineer at IBM?
Paul: Oh, definitely the focus of a distinguished engineer is to leave a portfolio of products in a particular domain. So in this particular case you know I have several product teams that work on different offerings in the DevOps space. And I’m responsible for crafting the technical strategy and providing architecture and technical direction to multiples of these teams.
Shimel: Got it. So CJ, as I think I mentioned to you in our off camera or off microphone conversations, you know the month of September is gonna see DevOps.com focusing a bit on APM and DevOps. And of course this is a subject that you are you know very familiar with. And part of what you do in Bluemix and all of these things, obviously APM’s a part of it. But why don’t we start with this? Can you define for our audience, when we talk about APM and DevOps and maybe it’s in relation to ITSM, right? IT Service Management which is something you’re also very familiar with. Can you kinda pull this all together for us?
Paul: Yeah. Definitely. Traditionally operations teams have always worried about application availability and performance. Where performance includes you know how the application might respond from different locations around the world and how the response time evolves or changes for different types of transactions. Now, as we look at DevOps and the role of the DevOps teams becoming more cross functional, a lot of the focus is shifting to how do we get the synthetic monitoring, the configuration of the monitoring and all of that to be handled much earlier in the development cycle. In fact, in an ideal world, just like our source code is written to the specs of the product, we would also create the associated operational configuration of the monitoring tool, as well as the creation of the synthetics scripts, along with the source code. And so in the ideal case you would have all that be checked into the same repo. And then as part of your continuous delivery process you’d want to make sure the operations tool is configured to monitor that particular build that is deployed in a particular environment.
So that’s the shift in thinking that we see a lot of the teams going through is how do we actually make this possible you know where all of the setup and configuration is treated as an extension to the source code of the product and worked through the lifecycle. Because it gives various advantages. Let’s say a new build comes out and you’re going to go deploy that new build in a particular environment or perhaps even introduce it to a small subset of your users in production. When you introduce that new build with a new feature, you want to monitor that new build or that new feature. And take all of that data and make decisions on whether you now ramp up to your full population. So what we see happening is in the new world of DevOps and especially on the cloud, you know monitoring has to go very much hand in hand with the features that the development team has built. And so it’s no longer just an after the fact operations thing to worry about. It is something you have to worry about as an integral part of the DevOps lifecycle. So that’s why I think this is a really exciting time. Especially the cloud enables application teams to do A/B testing and evaluate you know performance of new features, the response times of new features as it compares to their existing version of the application that is running.
Paul: And so all of the new capabilities you know that are the focus area in the application performance management space.
Shimel: Sure. Exciting times indeed. CJ, I would be remiss if I didn’t mention on this very topic that we actually have a webinar scheduled, I believe it’s Tuesday, September 20th at 1 p.m. Eastern Time with a colleague of yours at IBM, Arun Bilgiri.
Shimel: Are you familiar? Yeah.
Paul: Arun Biligiri.
Shimel: And Arun is gonna be doing this on Learn Why We Must Shift APM Left in the DevOps Lifecycle. And he’s gonna be also joined by Dan Kern, Solution Architect over at Prolifics. So you know if you’re listening to this and this is a topic of interest to you, not you CJ, I know you’re interested. But our audience, I’d highly recommend. You can probably register for this off of our DevOps.com. Go to the Webinar and Events page and you can register there. But it should be dead on on topic for this. So CJ, I wanted to focus back on though this idea of shifting APM left, right. So when we talk about DevOps and how we fit in things like security and QA, now APM, that seems to be a common theme of shifting these things left. Get them earlier into the development process, right. Because the earlier it’s in, hopefully the better it comes out in the end, right. I guess my question to you CJ, is can we just continue shifting everything left, right. You know what I mean? Is there a point of diminishing returns of shifting left?
Paul: Yeah. I think in this particular case especially, you know it’s something that is a good thing to think about right from up front. Especially as the velocity of deployments increases, we would expect the development teams to be more and more you know working the monitoring of the applications in conjunction with the code. Because you’re just kind of thought to have a throw it over the wall kind of syndrome. Because you know that would be too slow in this world to make it successful. So that’s why I think the shift left is almost an inevitable side effect of the need to go faster. And also with some of the shift to cloud, you know the way things are done are also different. So more and more the application architectures and topologies are becoming more complex to handle geographically distributed deployments and so on.
So more and more we expect the development teams that are involved to be doing some of these operations automation just like they would work on you know features for the product. So some of these concepts surround you know cycle liability engineering and so on that are being out in the industry are just examples of why you know we’ll get more and more of a DevOps perspective on things that used to be just purely operations activities in the past.
Shimel: Got it. CJ, my next question with you on this is and by the way we’re running low on time so I only have one or two more questions left. Is you mentioned earlier one of my favorite things happening in IBM and that’s the Bluemix Garage method, which is an open source methodology kind of taking best practices and lessons learned from IBM themselves, as well as your partners into this methodology. Talk to us a little bit. Where does APM fit into there?
Paul: Yeah. So with this Garage method what we’ve been doing is to take the learnings from across the industry on agile practices and design thinking and you know the latest DevOps practices and lean practices and so on. And organize it in such a way that enterprises can learn how to act like a startup but do it at enterprise scale. So within that method, we talk about how you think and plan, and code, and deliver, and run, and manage, and learn around a whole culture of DevOps. So in this particular case as we talk about application performance management, we typically think of it in managed. But as we have been discussing here, it kind of permeates through the lifecycle because you have to think about the availability of your applications as you design and build them. And you have to write the code for the synthetic monitoring steps, along with your features of the product. And you have to deploy them as part of delivery. And then clearly you’ll collect and analyze the data as part of manage and learn. So I think this whole shift left of APM fits in very nicely with the Garage method and its different lifecycle phases.
Shimel: Got it. CJ, I apologize, but just as I promised, yeah, we’re way over 12 minutes already and we’re gonna need to wrap things up here. But before we do, I wanted to just ask you one final question which is usually one that I ask most of our guests here on DevOps chats. CJ, for our listeners out here, if you had to recommend one book that they should read, what should be the next book they read?
Paul: I would highly recommend folks read the Site Reliability Engineering book that has come out of Google. They talk about how engineers with development skills can automate a lot of what used to be production activities. That’s a very good read. Within IBM we’ve been doing these kinds of activities ourselves. You know me make sure that the individual squads are also responsible for their performance. But I find this is a good book for teams to look at if they haven’t already read it.
Shimel: Got it. Okay. CJ Paul, Distinguished Engineer at IBM. Thank you for being this episodes guest on DevOps Chats. Look forward to seeing you in person at some conference or another soon, CJ. And until then, keep up the great work.
Paul: Thank you, Alan.
Shimel: This is Alan Shimel of DevOps.com for DevOps Chats. Thanks everyone. Have a great day.