Continuous Delivery Foundation cdCon 2021, 24 Jun 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Sponsored Keynote Session: Slicing and Dicing: Building DevOps Word Salads - Jeremy Meiss, CircleCI

Description

Sponsored Keynote Session: Slicing and Dicing: Building DevOps Word Salads - Jeremy Meiss, CircleCI

While it seems that every week new terms appear to describe DevOps tools, segments, ideas, practices, etc., are they really new? CI/CD, Progressive Delivery, AIOps, GitOps, TreeOps (not real, but who knows, maybe it will be next week?) - are they all the same, or are they just reimagining last week’s buzz words? In this talk we will break down a few of these terms, what they mean, how they’re used, and why they may matter for you and your teams.

For more Continuous Delivery Foundation content, check out our blog: https://cd.foundation/blog/

A

Hi, I'm jeremy and going to be talking about deploy, release cicd, oh my or slicing and dicing devops word solids all right. So while it seems that every week we have new terms that tend to appear on the you know, landscape, to describe devops tools, segments ideas, practices, etc. Are they really all that new? uh You have ci cd, progressive delivery, ai ops, git, ops, tree ops, okay, not real, but who knows- and maybe it will be next week- uh and probably you know as the side note- I literally registered devtreeops.com for this talk.

A

uh I have over 20 domains that are really doing nothing. uh Hi, I'm jeremy, meese. I have a problem. I am the director of devrel in community at circleci. We are a continuous integration, continuous delivery platform, uh and you can find me on the twitterverse at I am jairdog or on polywork at timeline.jaredog.me.

A

Okay. So, with all these terms that are out there, are they all the same or are they just kind of reimagining? Last week's buzzwords?

A

Sometimes it feels like companies they like to kind of chop up a lot of these buzzwords, throw them all together, uh create a new service, a new you know, bc back venture and then voila. We have a new devops salad, so, as we think about the devops life cycle, there's the hope and then there's the reality.

A

John ospal called it uh work as imagined versus work is done, and I think this image here kind of really illustrates this. uh There's that hope of of how we want things to go and then there's the reality that you know a lot of us fall into.

A

So I want to do is kind of start from with some of the more recent trendy terms and ideas or buzzwords and then finish with some more foundational elements.

A

So we'll start with aiops gartner coined the term aiops, which is short for artificial intelligence for it operations and defines it as aiops combines big data and machine learning to automate I.t operations, processes, including event, correlation anomaly, detection and causality determination.

A

Those all seem to be buzzwords in and of themselves, but their gartner. They can coin the term another way to put this was at wikipedia, and it says there is it's an industry category for machine learning, analytics technology that enhances it operations analytics in these tasks. Kind of mentioned here would be things like uh automation, performance monitoring and event correlations uh among, among others.

A

Now an ai apps platform has two main aspects: one is machine learning and the other is big data in order to collect data from the the big data platform for the purposes of observing it and engaging with it, you have to shift away from kind of that segregated I.t data towards a holistic machine learning and analytics strategy. The goal, then, would be to receive continuous insight, so you can continuously fix and improve via automation, in its simplest form, which this image looks. Really simple.

A

Aiops platform helps aggregate and make sense of disparate data sources, identifying the links and interrelations between different silos and then helps predict the events, prescribe solutions and then act based on real-time data streams.

A

All right, the next term get ops. So this one is relatively new and there isn't even a wikipedia page for it. It's just started appearing in the last few years and is actually newer than aiops and there's not even a page for git space ops. So I feel like this is a big oversight.

A

You know by the devops word solid gods to not have anything on wikipedia for get get ops or get space ops and show. uh This should probably be rectified soon. I'm sure someone out there is going to jump on that and make it a making a new thing now weave works themselves kind of describes, git ops as it's a way to do. Kubernetes cluster management in application, delivery, attempts to take existing continuous delivery and infrastructure as code principles which are non-tool specific and then repackage them all down into tool.

A

Specific activities so get ops kind of being this a tool specific thing: instead of the overarching, non-tool specific principles it kind of has these these activities would be, uh the entire system is described, declaratively uh the canonical system state is versioned in git and then approved changes can be automatically applied, and then software agents insert correctness and alert on any divergence that might be there now, as I've seen it over.

A

You know the last few years this term never ceases to carry its own weight and controversy on devops twitter, every so often uh right up there with you know whether html is code uh and and others, so uh you know, does for a great activity, go out and search, get ops and uh you'll you'll be treated to some interesting um debate.

A

Okay, so progressive delivery. Now this is a new term uh in new. You know it's two years, so it's that's kind of new in in this this kind of field. It was coined by james governor of redmonk. After hearing about microsoft's progressive experimentation model for feature rollouts, he wanted to capture a wide range of new software development practices that went beyond just continuous integration and continuous delivery, and so he said that you know. Progressive delivery is continuous delivery with fine grain control over the blast.

A

Radius, uh adam zuman, formerly vp, of product at launch darkly realized that feature management would be a key enabler of this new progressive delivery model. A model that's defined by dark launches, feature, flagging testing and production, canary launches blue green deployments, a b testing and so on, and so there are two tenants that kind of make up this progressive, progressive delivery methodology and they are release progression, and this refers to adjusting the number of users exposed to and impacted by new features at a pace, that's appropriate to your business.

A

You can perform release progressions in a variety of ways via targeted rollouts, canary launches ring deployments, percentage, rollouts and so on. You might also think of a release progression as a progressive deployment or progressive release.

A

The second piece is a progressive delegation, so this refers to progressively delegating the control of a feature to the owner who's most closely responsible for the outcome of that feature or of the overall product. A simple way to think about this is you know, initially engineers, they control a feature they're, the ones that are you know uh responsible for building it and testing it, and so that makes sense. But once the feature has been tested, uh it makes more sense for a product manager.

A

The pm, in this case for uh to you, know, go and control the process of launching that feature and then letting the engineers get back to engineering.

A

So ci cd- this is our next thing that we'll jump into refers to continuous integration and continuous delivery. We're going to separate those terms out a bit, so we get a little clearer definition here, so continuous integration, it's a practice that encourages developers to integrate their code into a main branch of a shared repository early and often so, instead of building out features in isolation and integrating them at the end of a development cycle code is integrated with the shared repository by each developer multiple times throughout the day.

A

Continuous integration is a key step to digital transformation, so continuous integration kind of the what it is is every developer commits daily to a shared main line. Every commit triggers an automated build and test, and then, if build and test fails, it's repaired quickly within minutes and cycle continues until everything passes and then the why of continuous integration would be like improve team productivity efficiency, happiness, find problems solve them quickly and then release higher quality, more stable products.

A

Now continuous delivery is the practice of ensuring that software is always ready to be deployed. Part of that is testing. Every change you've made so, in addition, you've also made that effort to package up the actual artifacts that are going to be deployed. Perhaps you've you know already already deployed those artifacts to a staging environment. So continuous delivery requires uh continuous integration, they go hand in hand. You can't have continuous delivery without you know continuous integration, relying on the same underlying principles for both of them that you're cutting work into small increments.

A

Now over the years, I've noticed that a lot of it professionals and vendors confuse the words, deploy and release and there's actually a big difference, and a lot of it professionals are just are not aware of this or you haven't really been exposed to it, and, uh and so I'm going to kind of break this down into how release deployer independent in crucial steps in the your process of shipping, uh shipping software shipping yeah so deploy a deployer deployment includes all of the technical activities that are needed to make a software system or feature available for use.

A

You can think of a docker container running in a pod on the kubernetes cluster. The piece of software passed all the tests in your ci cd pipeline and is ready to receive traffic from production users, but it's not actually receiving any of that traffic. Yet so this part of the process is just to make sure that the new version is healthy and running smoothly.

A

It takes care of all the technical checks and balances without any other risk incurred by actually serving production traffic. You can conclude that you know kind of deploying a piece of software is a mundane and relatively should be risk-free activity. You can get the software out. There can start to be. You know testing, uh and this is where you know. Maybe you might have some progressive pieces here uh to get it so that you know that it's ready with uh to go now deployments.

A

You know, are related to the overall continuous delivery process, but they tend to be more automated. Thus, you know getting them ready to be released. So when you're ready, it's it's ready to, you know, get kicked out there.

A

Now the release itself, a release, comes after a deployment and includes all of the activities that are needed to move part of or all production traffic to the new version, all of the risks uh and things that could go wrong, like downtime lost revenue, a bad customer experience are related to the release and not to the deploy to production.

A

Releasing software is a risky activity that needs to be done in a slow, controlled fashion. The release cycle is the overall process of deploying software to an environment and then ensuring that all of the deployments in a release are functioning and and performing as expected. So one could view release processes and release management as related to the continuous delivery concept, but that they're actually individual processes themselves.

A

This would be where practices like progressive delivery or release orchestration would come in assigning version tags, different segments of people, users that would be using it so on.

A

Okay, so, uh hopefully I've been able to break down some of these kind of overarching things that all kind of really relate to devops and the different word salads and touching on release and deploy and how they're different. I would love to have more conversation around this uh as you, uh you know, kind of think about what we've talked about. If you have any questions you can find me on twitter at I am jairdog. You can also find me on linkedin, timeline.jareddog.me and yeah. Thank you. I appreciate appreciate you coming to talk.