Mesos Is to the Datacenter as the Kernel Is to Linux


Necessity is the mother of invention. We needed our datacenters to be more automated, so we invented tools like Puppet and Chef. We needed easier application deployment, so we invented Docker. Of course it didn’t stop there. Ben Hindman, the founder and chief architect of Mesosphere, co-created Apache Mesos. In his keynote at MesosCon Asia 2016, Hindman relates how failures and elasticity led to the development of Mesos.

Hindman observes that the natural course of progress is solving old problems, and that creates new problems. Or perhaps a better way to think of it is new solutions create new opportunities. “As we fix some problems,” says Hindman, “We’re able to take on a new class of problems. So now, as devs, we said, “Hey. The machines that were running my Docker container or my app have failed. Can you figure out how to run this on different machines?” Or they said, “Hey, I’ve got a bunch more users right now, can you run my container on a whole bunch of different machines? Can you just scale it up with the click of a button?” So these two new problems–failures and elasticity–drove us to things like Mesos and Marathon.”

So then we have our checklists. Everyone has checklists, don’t they? We’re solving challenges and building new things. Hindman says, “You’ve got to figure out service discovery. You’ve got to figure out load balancing. You’ve got to figure out networking. You’ve got to figure out how you’re going to do storage volumes, security, secrets, health, metrics, logs, debugging, so forth and so on. Most organizations, they start here. They start with Mesos and Marathon and then what they find is over time, they need to start solving these other components, which are not core aspects of Mesos and Marathon themselves, for their own businesses.”

So the checklist grows, and items are checked off, and what we have is not just Mesos but a full-blown ecosystem, and then the core components of this ecosystem are bundled together as DC/OS, the datacenter operating system. DC/OS is a distributed operating system based on the Apache Mesos distributed systems kernel, and it manages your datacenter as though it were a single machine. “It’s this idea that Mesos,” says Hindman, “as this core component in the system really acts more like a kernel to the data center operating system in the same way that Linux is the kernel to CentOS or Ubuntu or Debian.”

Of course, this is not a stopping point because there is never a stopping point, and Hindman’s team is developing a DC/OS software development kit (SDK) to support the development of even more sophisticated distributed systems.

Watch the full keynote (below) to learn where Mesos and DC/OS are going, and how to be a part of their amazing progress.

Interested in speaking at MesosCon Asia on June 21 – 22? Submit your proposal by March 25, 2017. Submit now>>
Not interested in speaking but want to attend? readers can register now with the discount code, LINUXRD5, for 5% off the attendee registration price. Register now to save over $125!