Blog
7 June 2021
Oleg Zinovyev, technical writer

Why Flant acquired Okmeter monitoring company and what we should expect from this deal

Okmeter is a server monitoring solution that is available in both SaaS and on-prem versions. Flant and Okmeter have been working together since 2017. For Flant, Okmeter is one of the primary tools for monitoring customer infrastructure; over the years, companies have been cooperating to improve Okmeter observability capabilities.

PostgreSQL monitoring in Okmeter

In mid-May, Flant announced the acquisition of Okmeter. Perhaps, this deal came as a surprise to some participants of the DevOps market. However, in many respects, it is a natural consequence of fruitful cooperation between the two companies.

This interview with Nikolay Sivko, co-founder and now former owner of Okmeter, and Andrey Kolashtov, co-owner and managing partner of Flant, addresses many deal-related questions circulating in the community. Nikolay and Andrey share their views on the future of Okmeter and its coming features. They let us glimpse into decisions that have led to the deal and how it will affect existing and future Okmeter customers.

Nikolay Sivko speaks at the HighLoad++ conference

Some background

— What were the actual reasons for this deal?

Nikolay: It all started three or four years ago. Flant came to us and said: “You’re running way too fast; we are tired of catching you up. Moreover, monitoring is not our primary objective”.

Andrey: At that time, we were focused on the platform. First and foremost, we wanted to do our DevOps part well. We strived to provide services to our customers in the most effective way, thus we were looking for an efficient and ready-made monitoring solution. It was a strategic collaboration for both companies.

 

Nikolay: Right, it was natural cooperation. We benefited from Flant’s feedback and expertise. Flant needed a solution that “just works,” there was no opportunity to build it from scratch. It was a truly symbiotic relationship for both of us.

Andrey: We were exhausted from setting up and configuring various well-known monitoring systems over and over again, then attaching exporters to them in the hope of getting an efficient and universal monitoring solution. And we did not have a dedicated team to handle the essence of monitoring: what metric to monitor, how to implement it, how to visualize it, etc. Of course, we had our own monitoring solutions, our developments, but I cannot say they were perfect. Okmeter did just that.

 

— How did the idea of the deal come up?

Nikolay: Flant was our biggest customer. At some point, they felt the need to influence the product more and meet their specific needs more quickly. They wanted to have control over Okmeter. We liked this idea.

Andrey: Right. Well, the area of monitoring is crucial to a company that is engaged in infrastructure maintenance. Monitoring helps you to detect problems on the customers’ side promptly and prevent them. So we wanted to have our monitoring solution, and we wanted to control it fully – e.g., quickly add to it the required functionality or modify it.

From the business standpoint, it also allows Flant to generate product-based revenue. That is, to sell the product in addition to services. And this product will now contain even more of our expertise.

Andrey Kolashtov (center) at HighLoad++ Spring 2021

 

— Will Nikolay participate in the further development of Okmeter?

Andrey: We would like to benefit from Nikolay’s extensive experience while implementing some monitoring functions. We will definitely use his help and consultations at an early stage.

Nikolay: Currently, our team delegates all the responsibilities to Flant specialists. In doing so, we make sure that the service runs as expected and our customers are not affected. Also, we share our vision regarding the development of Okmeter. However, from now on, Flant will develop the product independently.

The thing is that Flant already has extensive experience with Okmeter and knows a lot about it. Thus, Okmeter is not a mysterious black box for them. Flant has its vision regarding the development of the service, and they have already begun to implement it. They can’t wait to do things in the right (their own) way.

How the deal will affect the market and customers

— What is the impact of this deal on the local monitoring market?

Andrey: The deal will bring to the market the real-life experience of experts who operate the servers.

Nikolay: I think the deal will mainly influence the DevOps market and related areas rather than the monitoring market. You see, the deal makes it possible to embed Flant’s service-related expertise in the product. And this expertise is, obviously, cheaper than hiring professionals in the area of monitoring. It means that, for some companies, Okmeter can even substitute those professionals.

 

Andrey: Okmeter is a monitoring service that does a lot of things automatically. It can automatically find, e.g., a Postgres database, display the related graphs, and send alerts if problems are detected. In other words, you do not have to create those alerts manually and add them. Instead, Okmeter will automatically detect what should be checked to keep your infrastructure healthy.

Nikolay: For example, there is one traditional and quite popular monitoring solution. Everyone knows that this thing cannot do anything automatically after installation. Essentially, it is useless until you manually configure it. It is just a tool, and you need an experienced specialist to use it in a meaningful way.

At the same time, Okmeter provides a solution that works automatically and out-of-the-box. It is similar to the human expert looking at the graphs and saying: “Look, the problem is here.”

 

— What about the existing Okmeter customers? How will the sale of the service affect them?

Nikolay: Obviously, it will affect them positively. Flant has the expertise that Okmeter didn’t have, and customers will now benefit from it. By using all their knowledge in monitoring, Flant can actually make it way more awesome and impressive.

Andrey: From the economic side, we have no plans to change tariff plans. From the technical side, the impact of the deal is designed to be positive. We will further develop the product, add new features for it, and do it very actively — currently, we are recruiting a large team to develop Okmeter.

The deal and the OVH incident

— There were rumors that the sale of Okmeter is related to the recent fire in the data centers of OVH. Back then, the Okmeter monitoring experienced an outage for a while. Is this really the case?

Nikolay: No, the deal just coincided with the fire; the incident did not affect our decision to sell. However, generally speaking, it was an unpleasant situation for Okmeter and our customers. Thanks to the timely and comprehensive communication, only a few clients stepped out. Therefore, this incident did not cause any substantial damage neither to Okmeter’s reputation nor business performance.

 

OVH data centers devastated by fire, March 2021; photo by Xavier Garreau

 

— What was the problem with OVH?

Nikolay: The thing is that OVH didn’t really share detailed information about its data centers. We believed the three data centers in Strasbourg that hosted the Okmeter infrastructure constituted a fully-fledged availability zone (meaning that servers were independent). So we relied on it, and that was our biggest mistake.

We see the only solution to this problem — to use different providers in various regions to ensure this situation will never happen again. And, as far as I know, Flant is actively working to address this issue. Obviously, if it weren’t for sale, we would have to do it ourselves.

 

NB: Soon, we plan to publish an article about this fire and what Flant engineers did to minimize its consequences.

Future plans

— What are your plans for the service?

Andrey: We see Okmeter as several complementary products: storage, platform, and insights.

As the name suggests, the storage will store metrics, alerts, logs, and traces efficiently and cost-effectively. It will be able to store both Okmeter metrics and metrics of third-party monitoring services. E.g., users will be able to connect Grafana and browse metrics kept in our storage.

The platform will provide a user-friendly interface for viewing graphs, alerts and managing them. We will invest all our experience in the product to make an excellent tool for building dashboards and graphs, configuring alerts and sending them anywhere, and proper handling of incidents.

Insights are about ready-made metrics, dashboards, and alerts for popular tech stacks and use cases. We plan to create convenient tooling that is helpful out-of-the-box. In some sense, the insights team will be dealing with the “meaning,” with the essence of monitoring. This team’s mission is to figure out what parameters to monitor, what is happening in the services, whether they are running as expected or if they are on the brink of a failure. This team will also be engaged in building highly informative and convenient dashboards. Thus, the user can get the idea of what (s)he needs to do to eliminate the problem via a simple glance at the dashboard (without poring over the logs). That is, to speed up the process of removing the problem as much as possible.

 

— Tell us more about the platform and the Okmeter interface in particular: what will be improved first?

Andrey: We plan to improve the visualization of specific components (depending on our and customers’ needs). We will also improve integration with the Open Source monitoring services, add the ability to connect third-party plugins to Okmeter, add to the interface graphs and dashboards derived from other instruments. In fact, Okmeter already has all the essential graphs and components required for monitoring; the thing is, we need some particular features, and we are going to implement them.

 

— Will developers only form the teams?

Andrey: Yes, the storage and platform teams will only include developers. The insights team will consist of developers as well as experienced SRE engineers. The latter will try to figure out what has failed on the client’s side and why we could not anticipate that. Their other objective is to define what parameters need to be monitored to prevent such problems in the future. A large team will handle both interface and agent (the custom Okmeter client) improvements and accelerate the troubleshooting.

 

— So they will be operating engineers?

Andrey: Not really. They will be involved in incidents but will not be dealing with their consequences. Instead, they will be analyzing the situation to figure out why Okmeter did not predict the failure. For example, suppose there was a database failure. Its buffer had been filling up for some time. We missed this process on our graphs; there were no alerts, although we could have known about the buffer filling up a week earlier. Here, the insights team comes into play. It finds a way to monitor this buffer and help the database administrators, thus preventing potential database failures in the future.

 

— Do you plan to increase the number of supported technologies, e.g., databases?

Andrey: Of course. Our plans are very ambitious. We intend to add to Okmeter support for many popular technologies: MongoDB, ClickHouse, ProxySQL, HAProxy, Ceph, etc., and expand the functionality of the existing integrations. We will also significantly expand the Kubernetes monitoring.

 

— Flant has broad expertise in Kubernetes monitoring. How do you combine it with Okmeter?

Andrey: Yes, we already monitor Kubernetes using our custom-built tools. Our monitoring is based on Prometheus and a bunch of custom third-party and in-house exporters. For these tasks, we have a dedicated Flant team responsible for the Kubernetes platform. We have put a great deal of work into proper Kubernetes monitoring.

At the same time, Okmeter also integrates well with Kubernetes. You can already use Okmeter with Kubernetes in the same way as you do with other services and software on regular nodes. That is, you can install it in the Kubernetes cluster and Okmeter will automatically detect all the software in the cluster and try to connect to it to collect metrics. We will combine this integration and our tools/ideas into a powerful and versatile Okmeter-based solution, adding our dashboards, exporters, and the general experience to it in the process.

 

Example of our existing traffic dashboard for a specific namespace in the Kubernetes cluster

 

— And as for on-premises installations, what are your plans in this regard?

Andrey: We will foster this area as well. Okmeter used to be primarily cloud-based, but now you can install it as an on-premises component using Flant tools. We have already accumulated experience with such Okmeter installations.

 

— So there will be two editions?

Andrey: Yes, you can select the cloud or on-premises version. The cloud version is optimal if your needs boil down to monitoring, say, a couple of servers, and your security requirements are somewhat relaxed. On the other hand, if requirements in information security are more strict, you can install the on-premises version in your isolated environment. The operating principle is the same: an agent is installed on your servers and sends data to local storage instead of a cloud-based one.

 

— Do you plan to create Open Source components for Okmeter?

Andrey: Yes. The Okmeter storage will likely be based on Open Source components. We will share them via GitHub, and we will contribute to them, adding features that will help improve the storage. For now, we do not plan to make the source code of the platform and Insights freely available, but things may change. Obviously, that does not apply to components that are already Open Source. For these, we will bring our improvements to the upstream.

 

— Nikolay mentioned plans to improve the availability of the Okmeter infrastructure. How exactly are you going to do that?

Andrey: Well, this is one of our top priorities. We place particular emphasis on making highly available storage.

First, we want to make sure that within 30 minutes of the failure of the main data center, Okmeter will be up and running in the secondary data center — with all the dashboards, graphs, alerts, settings properly configured. That is, all the parameters will be monitored almost without interruption, while metrics will be pulled from the backup. To do this, we will use orchestration scripts that allow you to create a new infrastructure and deploy all the required software components all at once.

The second step is to distribute the infrastructure between different data centers in various countries. Then, customers will be able to choose the HA level (how many data copies to keep and in which data centers).

Global plans

— What is your strategy for increasing Okmeter’s competitiveness in the global market?

Andrey: First, we are going to enhance functionality greatly. Significant improvements will be made to UX to boost the user-friendliness and functionality of the platform.

Our second objective is active marketing on all fronts. Okmeter already has some efforts here, but we will significantly boost them.

We will also be looking for partners who can help us monitor specific services — for example, databases. We are interested in companies that have a deep knowledge of particular software. In this regard, we are open to cooperation.

P.S.

Learn more about Okmeter features here.