Moving to give IT organizations more confidence in an “open core” implementation of the Mesosphere Data Center Operating System (DCOS), Mesosphere this week announced the formation of alliance of vendors committed to developing DC/OS as an open source project.
Matt Trifero, chief marketing officer for Mesosphere, says that while Mesosphere will continued to sell its own implementation of DCOS the DC/OS will be provide an open score instance of its operating system that multiple vendors will contribute to on an ongoing basis. Thus far 60 organizations, including Accenture, Cisco, Hewlett-Packard Enterprise, Equinix and Microsoft, have committed to the DC/OS project.
Longer term, Trifero says it’s conceivable that organizations might even one day see multiple distributions of DC/OS; with the open source project guaranteeing workload compatibility across them. In the meantime, the Apache Mesos project will continue to be responsible for the core distributed kernel on which DC/OS and Mesosphere DCOS will both be based. Mesos itself can only be invoked using an application programming interface. The software applications wrapped around it are what make up the operating system built on top of that kernel.
As an open source project Mesosphere will have more appeal to IT organizations that are committed to preferring open source software whenever possible. At the moment, Mesosphere DCOS tends to appeal to higher end IT organizations that have a lot of internal engineering expertise. Via the efforts of all the vendors contributing code to DC/OS, Trifero says one of the goals is to make DC/OS more accessible to the average IT administrator.
Coinciding with the launch of DC/OS, Microsoft this week also revealed that its Azure Container Service now supports DC/OS.
Ultimately, DC/OS providers a framework that make deploying complex distributed systems, including HDFS, Apache Spark, Apache Kafka, Apache Cassandra and others, as DC/OS service using a single click. It’s up to each IT organization to decide what orchestration framework they want to then plug into DC/OS. In fact, IT organizations will be able to opt to deploy all or DC/OS or just some of the 30 plus modules that make it up. In any case, the way modern container applications are provisioned, deployed and managed will become a lot more automated.
Naturally, there is no shortage of container management platforms these days. The degree to which IT organizations standardize on one versus another obviously remains to be seen. For that matter, many vendors appears to betting on multiple management frameworks that to one degree or another might be in contention with each other in terms of how a specific management function gets addressed.
But in the case of DC/OS the goal is obviously to leverage automation to eliminate all the tactical deployment issues associated with deploying an application on top of a container. The level to which that level automation alleviates all DevOps issues is still largely unknown. But it’s clear that going forward the level of manual effort associated with deploying applications across the enterprise will be sharply reduced on a platform such as DC/OS.