Containers promise to be the next big thing for enterprise infrastructure. Apache Mesos is probably the most ambitious project in this area: it not only tries to solve the problem of scheduling containers on a cluster of machines, but aims to do so in a fully fault-tolerant, scalable and modular way. With Mesosphere DC/OS there is a product-quality version of Apache Mesos available and even better: it is open source now! This is great news as it will be a big boost for the Mesos ecosystem.
Quobyte’s Mesos framework allows deploying a POSIX distributed file system to your cluster in just few simple steps. And being able to use DC/OS everywhere, the setup process becomes quite a bit simpler, yet again.
What Quobyte does, is solve an essential problem with containers: running stateful applications in a container infrastructure. All data in Quobyte is accessible from all hosts and containers concurrently. And since Quobyte is a parallel file system with high-performance block access, you can run all sorts of applications from Big Data to databases like MySQL – without compromise. And if you need to ensure exclusive access (think: database), Quobyte’s automatic locking mechanisms has you covered.
We at Quobyte were very happy upon discovering the Mesos project a few years ago. A container infrastructure with Mesos’ features complements Quobyte’s capabilitiesperfectly. Together with Mesosphere’s DC/OS the two greatly facilitate infrastructure management. Quobyte does for storage what Mesos brings to compute: it turns servers into a fault-tolerant and scalable general-purpose infrastructure. Note that that’s the kind of infrastructure that Google pioneered in their data centers for their custom software stack.
This new kind of infrastructure fundamentally changes how we use and operate infrastructure. With the separation of the application from its execution environment, application teams can consume infrastructure in a self-service manner without having to coordinate themselves with the infrastructure group. By decoupling infrastructure from individual pieces of hardware, infrastructure teams no longer need to bother hardware ops routinely at night. And the hardware team does not need to coordinate with infrastructure and applications for doing repairs and maintenance. Everyone is more productive and happy, and much larger infrastructures can be operated by a small group of people this way.
Check out Mesosphere DC/OS at: dcos.io.