Is Apache Mesos dead, and why should you care
Apache Mesos is a distributed kernel that enables multiple machines to work as if they were a single computer. Mesos is primarily used in Big Data deployments alongside Hadoop, Spark, Kafka and others. But is Mesos still relevant or is Mesos dead? How well does it compete with Kubernetes? Let’s try to answer this in 2020.
Mesos: a rapid decline
Google Trends leaves no doubt: Kubernetes is on the rise, while Mesos and DC/OS, one of the most prominent software built upon it, are declining. At the time of writing this article StackOverflow has 22609 questions tagged Kubernetes, while Mesos only gets 1225. It may be safe to assume Mesos is dead, but it may not be the whole truth.
Should you really compare Mesos to Kubernetes
There are many features overlapping between the two, however you should not forget that Mesos isn’t only about containers. Mesos was born to achieve a single goal: make a cluster of computers look like a single computer. Be it 100 machines or one single laptop, Mesos enables applications to be run on both. Kubernetes is aware of the number of nodes and it uses them as such: nodes. Engineers can decide to schedule certain workloads to certain nodes (using taints and tolerations). On top of that Mesos doesn’t pack a scheduler while Kubernetes does.
Don’t forget that Kubernetes is the way Google manages and scales cluster and applications, Mesos is different take at it with different objectives and requirements.
In the end, is Mesos dead?
While Mesos is nowhere near Kubernetes, its development is still active but it lacks manpower. Most of the market has been literally gobbled down by Kubernetes hence all the interest and new shiny things flourish near it rather than near Mesos. You can actually see that according to Google Trends, Mesos has been surpassed by Docker Swarm which was put in the chart for reference.