D2DO313: Operations is a Reconciliation Loop (Sponsored)

Kyler Middleton (0:07)

Welcome to Day 2 DevOps for the Dev-Oops is in the details. I am Kyler Middleton, and I'm joined by my user-friendly co-host Ned Belavance.

Ned Belavance (0:17)

I am very user-friendly. Hey, Kyler. Today, we're discussing the reconciler pattern for operations. The place of non-determinism in your deterministic deployment process and what you can do about it, namely, attend the Gartner's December IOCS Conference in Las Vegas.

Kyler Middleton (0:35)

With no further ado, let's get to it. Paul, thank you so much for joining us. We would love to hear about what your experience is and what your role is at Gartner.

Paul Delory (0:46)

Thank you very much for having me, first of all. Always appreciate talking to some very smart people about some very pressing issues in the industry.

But yeah, my name is Paul Delory. I am a research vice president here with the Gartner for Technical Professionals Research Organization. And that means that my research is really aimed at engineers and architects. So my remit here at the firm is to kind of get into the nuts and bolts details of how are we actually going to do all of these lovely things that we want to do, right?

So I do cover Infrastructure, Automation and DevOps. That's where I spend most of my time. I also spend a fair bit of time though talking about private cloud architectures. And in particular, these days, I spent a lot of time talking about VMware and VMware alternatives.

That is my background. I came up as a virtualization engineer running VMware and Citrix. And, you know, but I always had that special interest in automation. My attitude was, if I'm going to do something more than twice, I'm writing a script, right? I don't want to do click offs. I don't want to do any of that. So that's really where I enjoy spending my time and talking through some of those issues with our clients.

Ned Belavance (2:19)

That's great. The first time that I started using virtualization and realized that there was a PowerShell module for VMware, I was like, I am just going full bore at that. I don't want to touch the vCenter console if I don't have to.

Paul Delory (2:35)

Like one of the best PowerShell implementations too.

Ned Belavance (2:39)

Yeah, for sure.

Paul Delory (2:40)

I learned PowerShell mostly to be able to use PowerCLI. I mean, and I think that's true of a lot of VMware admins.

Ned Belavance (2:47)

There's that and Exchange 2007 You could only do certain things through PowerShell. So between those, we're dating ourselves, aren't we?

Let's talk about something a little more modern. DevOps. That was a practice that came around that general timeframe and it has continued ever since. It's given us things like continuous delivery and continuous deployment. But when we were talking on the prep call, you mentioned you think there's some things that are still missing from that picture.

Paul Delory (3:20)

Absolutely. I think everybody agrees that they're a third phase. It goes continuous integration, continuous delivery, and then something else.

Because the definition of continuous delivery, the definition of done is that you have an artifact. Whether that's a container or that's an OS image, or if you're doing infrastructure as code, maybe that's an Ansible playbook that you have produced out of a continuous delivery pipeline. But you have some kind of artifact that is ready to be deployed. But it hasn't yet been deployed. And you haven't done all of the operational things that you need to do.

So there does have to be this phase, this next phase. And for me, as an ops guy, that's kind of where my job starts. So if you think about a developer has produced a container or some kind of artifact, my job is now to deploy and operate that container. So it may be that the developer kind of stops thinking about it after continuous delivery, but that's kind of where I start thinking about it. As an ops guy. So that's the area that's really important to me, and that's what I'm exploring. So the focus of my research right now is what I want to call continuous operations.

And that's basically extending the pipeline. I love continuous delivery. I love this idea of a fully automated, staged delivery pipeline where there is a test suite at every stage.

And I want to extend that pipeline. I want to carry forward essentially the ethos of continuous delivery into operations. I want to have the same automation and the same sequencing and the same formal verification. But I want to do that for everything that operations does, not just deployments. Because the picture is a lot broader than that.