Explanation:
Azat Khadiev: Hello. My name is Azat Khadiev. I am a PaaS developer for Mail.ru Cloud Solutions. Pavel Selivanov from Southbridge is here with me. We are at the DevOpsDays conference. He will give a talk here that you can build DevOps with Kubernetes, but most likely you will not succeed. Why such a dark theme?
Pavel Selivanov: It's actually not gloomy. It is about how many problems in our community we are trying to solve with the help of technology. And moreover, we try to solve with the help of technology rather one-sidedly. Kubenetes is the same - this is the thing they are responsible for, we can say that Ops. But we have a wonderful concept of DevOps engineer. A DevOps engineer is responsible for Kubernetes. At the same time ... Like you make Kubernetes, and the Dev guys are not aware of all these Kubernetes at all, they are not aware of what it allows you to do - and everything lives for them in exactly the same way. And this is despite the fact that Kubernetes contains ready-made solutions, ready-made tools in order to use this technology to stretch this DevOps approach, communication between Dev and Ops. We use this opportunity very little. Due to the fact that we are even transferring the current structures to all these DevOps tools - Docker, Kubernetes, clouds, and so on - we exacerbate this situation even more. And the tools are beginning to be used not as they were intended. And just terrible crutches are being built around all these technologies.
Azat Khadiev: I understand. The topic seems to be broad. What do you think is the most common problem companies have right now? With Kubernetes.
Pavel Selivanov: The most common problem with Kubernetes is the lack of competencies. This is a common problem in IT. There is always a shortage of specialists. There is always a lack of skills. And now with Kubernetes there are not enough competencies. And at the same time, there are still XNUMX% ready-made solutions on the market that would allow you to get Kubernetes, but at the same time not have the necessary competencies, there are still frankly few of them on the market. And those that are, they all raise some questions. Constantly with Kubernetes, we are in search of people who understand this. We are trying to pull the development under it.
Azat Khadiev: And given the current shortage of personnel in IT. Which has always been. And now there is. How do you think, how to live in these conditions? What are life hacks?
Pavel Selivanov: Life hacks. Firstly, from the point of view of the clouds, the life hack looks like this - and let's give us some of your competencies. And we will take them. And we will do it within ourselves. And that's all good. Except for the fact that it is important to understand for those who use it ... Actually, a great moment ... But it is important to understand that if you give part of your competencies somewhere to the cloud or provider, we will receive a universalized solution in return. To put it bluntly, we have a database that does very specific things, and it's been set up in a very specific way. Giving this database to the cloud, we, of course, can fire the administrator who used to deal with database clusters - the same Amazon or Google will do it for us. But at the same time, the same Amazon or Google will not allow us to fine-tune our database. Large projects, large companies - in any case, they end up using cloud solutions at some stage of their lives, and then, in any case, they return to taking competencies back to themselves, because more specific is required.
Azat Khadiev: Are universal solutions bad or can more be built on their basis?
Pavel Selivanov: No, universal solutions are definitely not bad. Universal solutions are good. Just one-size-fits-all solutions…one-size-fits-all. It's important to understand here. It's like taking a common script... If you can build the entire logic of the company's work around this common script, then it's cool. And if the logic of work is different, but you take this universal solution, a universal script - and start pulling an owl on a globe, this is bad. There is nothing wrong with universalism itself.
Azat Khadiev: If this admin is already working for you, the point is not in his dismissal. He just can do more.
Pavel Selivanov: Yes, to take away from him the routine of giving them somewhere to someone to be made somewhere there. This is definitely a good approach. An important point here is whether this standard solution is suitable for a specific case.
Azat Khadiev: Just based on my experience, I see that many companies are doing the same thing. They set up a Kubernetes cluster, think about scaling it. And all these operations are very repeatable.
Pavel Selivanov: Yes, definitely. Moreover, if we take Kubernetes specifically, there is such a moment that there is really little deep, good knowledge on Kubernetes now on the market. And Kubernetes is such a gigantic constructor that if you took him to the company, be prepared to take an engineer with him who will do all this at full time. And it's expensive. And try to find such an engineer. If I talk about myself, I don’t really like some kind of cloud solutions, because I have a fairly good and deep understanding of how Kubernetes works. And often in the clouds I just lack some functionality that I request - and they tell me “No, you can’t”. Well, in that case, I'm sorry, but I can do better than cloud. But at the same time, if you don’t have a full-time engineer, you don’t want to pay for this engineer who pulls Kubernetes, and you constantly pay him a lot of money to just experiment, then the cloud is just a good great solution. Because at least there are guys sitting there whom the provider himself has already cheated. And they know what they're doing. And those basic things that you need on an everyday basis are there.
Azat Khadiev: What do you think about the current state of Kubernetes? What will happen to him in five and ten years?
Pavel Selivanov: Good question. I just know what's going on in the community about this. Some people believe that there will be nothing left but Kubernetes. The situation that happened a long time ago with Linux. That is, there are people outside of Linux who live on BSD, most likely they have very specific tasks. There are people who work under Windows - windows servers - most likely, they also have specific tasks, or they simply have competence in this matter and they are not ready to leave from there. In any case, the standard in our field is Linux. There is an opinion that Kubernetes will become the same de facto standard, and there will be nothing except Kubernetes. Kubernetes will manage not only applications, their deployment, deployment, scaling. Basically manage everything. Now they are already asking: “Is it possible to push a database into Kubernetes?” I usually say that the issue here is not Kubernetes, but Docker. If you are prepared that your database will run in containers, how will it work. They answer me: “No, no, no, wait. Containers are not required. Needed in Kubernetes. We will graft it to the node. That is, everything will be as we have it now, only Kubernetes will manage all this. And it's a good idea actually. That is, Kubernetes is such a thing when you can come to a company, if the company has Kubernetes and processes built on it, then a person who understands this - it is enough for him to watch a couple of days to say: “I am ready to support you. Fully. Whole. I understand how things work for you." Unlike approaches without Kubernetes, here some crutches were stuffed, here other crutches. Ansible here, Terraform here. Someone wrote all this, and it takes six months to figure it out. Here. So whether Kubernetes will become the de facto standard, I don't know. Today, he looks much more ambitious and confident than the solutions that are around him.
Azat Khadiev: Well, the comparison with Linux is quite bold. He works on one machine - and that's it. And Kubernetes runs on many machines. Immediately there is a million variations, causes. Yes, it's bold. Just considering that this paradigm has competitors. For example Serverless. Is kubernetes in danger with such competitors?
Pavel Selivanov: From Serverless… (laughs) Serverless — after all, we understand that there are servers. I recently heard a report on this. There, the person said that there are still servers - and this is a cloud. But we must always understand that the cloud also has servers. There are real iron servers, a rack, and now they are installed somewhere. Here is the cloud. On top of that, there is Serverless, where there are "no" servers. Therefore, the question is will Serverless Kubernetes win? It seems to me that Serverless will go to Kubernetes. For providers that provide Serverless, Kubernetes is a very convenient platform to provide it. Yes, perhaps at some point we will stop talking about Kubernetes in principle, as about ordinary business application development. But somewhere in the depths, providers and engineers will have Kubernetes, where all this will be implemented.
Azat Khadiev: A slightly different topic. There is such a thing as a fullstack engineer. What do you think of them? Do they even exist?
Pavel Selivanov: Um… Fullstack engineer… Well, it seems to me that it is worth delimiting these things about what… You know, there is such a thing as T-shaped people. Are such people needed in today's industry? Yes, definitely needed. We need people who have a broad outlook, but at the same time they are specialists in some narrow area of their own. And here the Fullstack engineer is the same - the person who does everything. Starting from frontend development, testing, backend, servers and everything else. I do not believe that in a large company one person can do this without having narrow specializations in each of the parameters. But at the same time, just having a narrow specialization, like what’s going on around this, I don’t know anything - this also doesn’t work in the modern world. That is, here I would say ... I would discard the word Fullstack. We need engineers very much. We need DevOps. I have a feeling we will revisit this soon. And they won't be needed.
Azat Khadiev: Can you reveal?
Pavel Selivanov: It seems to me that we in the industry will come to the conclusion that these roles of Dev and Ops will disappear soon. If we need specialists and we want to… We need such and such a developer, we need such and such administrators, we need DevOps engineers - now we have them, now we will have production engineers, SRE engineers. Although, in fact, what we need is the engineers we want to hire. The background is largely unimportant. Because... For example, SRE says that infrastructure problems are always software. And what ... Let's take developers - from the point of view that a developer is an engineer - put them in the support department and they will solve these problems in the same way as they solve business problems using code, using engineering, as such.
Azat Khadiev: And from this point of view… How to interview such engineers?
Pavel Selivanov: Oh, good question. He is probably already beyond what I understand in this life. But I would just give an example. It has nothing to do with the interview. This is about our education system in Russia. In IT, we know that our education system in Russia for the IT world is very outdated, it is not what it should be. I'm talking about the vast Russia on the average and what's going on there. People are graduating who are absolutely unprepared to go into web development, into a technology company the very next day after graduation. And it's kind of bad. We teach them some strange things, although it seems that we should teach them how to develop them for Android, for iOS, how to use Git and all these things. In fact, it seems like it doesn't. Institute is a time when your parents mostly pay for you. For your whole life. And you can dedicate five years of your life to studying in depth. And study all this T-shaped. When you can study at the institute what a version control system is, what development patterns are, how to test this whole thing, what databases and balancers are. And when you are already going to work, you begin to delve into a particular area. And that's how we get engineers. And our education system in Russia is much closer to this truth than we think. We are given a good mathematical training, we are given a good algorithmic training, we are given some idea about programming languages. And about the interview, it seems to me something close to this. We need to interview engineers. We want the top of the T on the T-shaped. Because he will acquire the vertical line of the letter T.
Azat Khadiev: Yes, interesting. Five years after the institute, it seemed to me that my education was somehow strange and inadequate. And then, in the course of work, when the tasks became deeper, the projects were larger, I realized that no, I was taught very important things. Pavel, thank you. It was very interesting to hear your answers. Let's hear your report.
Pavel Selivanov: Thank you.
Source: habr.com