Without a single "Oops": the top 10 reports DevOops 2017
If you have not been to DevOops 2017, then, probably, your thoughts are now approximately like this: “What does the Greek togas mean? What was the best presentation of the devops conference? What is this report and why is it recognized as the best? Were there any serious performances in ordinary clothes? ”
Serious was more than enough: if you are interested in the nuances of working with Kubernetes, Istio or Zipkin, in the post you will find all this, and the togs do not interfere with speaking sensible things. The title of “best” depended on the audience: we compiled this top 10 based on their ratings. And the answer to the first question is under the cut.
(keyout) DevOps in scale: Greek tragedy in three acts
Speaker: Baruh Sadogursky and Leonid Igolnik
Presentation of the report
In this case, it is important to understand the context. End of the day. The audience has already watched six reports, this closing keynote remains. Immediately after it will be a party, and after a whole day of active absorption of information, many probably want to go to it as soon as possible.
And something hardcore at this point would have been difficult to perceive - but the ironic costumed performance about devops, styled as a Greek tragedy, perfectly fit into the context, animating the entire hall. At the same time, “a fairy tale is a lie, but a hint in it”: the story of how a fictitious company faced various difficulties as it grew, helped not only to laugh, but also to reflect.
As a result, this very performance turned out to be the most highly rated by DevOops viewers. To share their feelings, we recommend watching the video in similar conditions: in the evening of a difficult day, with a glass of something you love.
Extend k8s
Speaker: Nikolay Ryzhikov
Presentation of the report
And immediately after the previous paragraph, there is as sharp a contrast with it as possible. In the report, who took second place, instead of jokes and entertainment, sheer technical specifics.
Kubernetes has become a key tool of our time (GitLab won because of him moving from Azure to GCP). And while some are already using it with might and main, others are eyeing it with interest. We recommend those who are “eyeing” the beginning of the report with introductory information, and already using it will be interesting further: Nikolai says “how everything is inside” and “how to simplify ops for developers with the help of k8s” (ideally, he would like would come to noops).
one-cloud: OS-level data center in Odnoklassniki
Speaker: Oleg Anastasyev
Presentation of the report
And who now, with all the popularity of Kubernetes, there are reasons not to use it? In Odnoklassniki: looking at it, they came to the conclusion "it is easier for us to wash our own than to adapt it for ourselves." But they took Docker there. All this happened within the framework of “creating your own cloud”: previously, the social networks used the principle “each server is occupied with only one task”, however, over time, they decided to perform different tasks on one server. And the report told in detail about this.
Will Odnoklassniki's experience be useful to others, if not everyone works in a company with several own data centers, and most of them have completely different specifics? Let's say this, it is unlikely that you will see the record with the feeling “we have to do the same thing,” but something from the report may be useful in a different context, and knowing “what other people live” is useful for a common understanding of the industry.
(Keynout) Come scale away with me:
Speaker: Corey Quinn
Presentation of the report
Corey Quinn is known as the creator of the weekly Amazon AWS newsletter Last Week in AWS . In his texts, he constantly sneers, dramatically different from the usual dry style of IT-materials. But with all the jokes, his newsletter was so informative and useful that even Amazon employees often follow the news on it.
Corey opened the conference with his keynote, and in his speech there was the same combination of “ironic and business”: one could hear about situations that are both laughter and devops sin. “I once sat on the Netflix report, where they said that they give all developers access to production in general. A man sitting next to me warmly approved these words. I looked at his badge and was horrified to find that he was working in a bank. ”
Silicon chainsaw massacre (as I spent the weekend on duty)
Speakers: Baruh Sadogursky and Leonid Igolnik
Presentation of the report
Again Baruch Sadogursky and Leonid Igolnik, but without the Greek and with a report on how to properly build the process of duty (and how it is often built instead). There is no longer a continuous “DevOps in scale” fun, but a more applied report with concrete conclusions. But if you know Baruch, you already understand that the place of the show was found here. First, the speakers do not simply list the theses “how to do it right”, but cite hypothetical stories as an example - and you can worry about her characters, recognizing yourself and others in them. Secondly, all this is illustrated with gifs - and on the videotape you can hear from the audience’s reaction how accurate their selection turned out to be.
Troubleshooting & debugging production applications in Kubernetes (aka The Failing Demo Talk)
Speakers: Baruh Sadogursky & Ray Tsang
And again Baruch, but this time with another speaker (and, as you might guess, in a different language).
With presentations in which there is a “live demo,” there is always the risk that “this demo on stage may not suddenly work.” And here they beat this complexity, deliberately making the demo “as if they suddenly refused” - and then, step by step, figuring out what needs to be done to make everything good. What goes well with the topic of “troubleshooting & debugging”: with debugging, it is interesting to understand a specific visual example.
Success story, or “Dev + DevOps + Ops”
Speakers: Artyom Kalichkin and Sergey Berdnikov
Presentation of the report It is
easy to extol the advantages of devops in conversations. It is much harder to really go to them when there is a large company with established processes that cannot change instantly. And if there are also circumstances like “this is a FINTECH, so safety is paramount”, the task becomes even more complicated.
And here is a report from those who have already passed along this path, where they share their experiences in detail, moving from “was” to “became”. It turns out, among other things, the story that for a devops one cannot simply take it and go to other tools by the decree from above - changes must occur to people and come from them.
Managing your microservices with Kubernetes and Istio
Speaker: Craig Box
Presentation of the report
Although the previous paragraph reminds us that devops are not reduced to tools, of course, tools and the ability to use them are important. And this report is strictly about the use of two specific technologies. If about Kubernetes is already spoken and negotiated (take at least the second item of this list), then here with Istio the situation is different: there are already over 9000 stars on GitHub , but there are not so many materials about it. The more valuable this report is from a Google employee - a company that is directly related to Kubernetes and Istio. If earlier they didn’t pay to Istio, now is the right moment: version 1.0 is about to be released.
How do you blame latency?
Speaker: Adrian Cole
Presentation of the report
Another technology report that helps to use the tool effectively. How do logging, metrics and tracing correspond? How to properly deal with the third? Why is the Zipkin tool so called?
The third answer would fall short: when Twitter suffered from falls and showed the so-called “fail whale”, the company called the tool created there by the word “harpoon”, hoping with its help to defeat the “whale”. And for answers to the other two questions, see the report.
From sysadmin to man
Speaker: Alexander Titov
Presentation of the report
Alexander is the managing partner at Express 42, which “grows DevOps in technology companies.” With this kind of work, it’s not surprising that he has seen many approaches to DevOps, including many unsuccessful ones, and he can summarize what problems often turn out to be and what we should do with them. What happens when devops comes exclusively from system administrators? And when only from the developers? And from the business that heard the buzzword and said “make devops yesterday”?
If you liked these recordings, you will enjoy even more at the DevOops 2018 conference: there you can not only hear the reports, but also ask any speaker in the discussion area. The conference will be held in St. Petersburg on October 14, we have already published a detailed announcement on Habré , and in the meantime tickets are gradually becoming more expensive - so it’s worth thinking about them now.