Mesosphere Intros SDK for Data Center Operating System

Aiming to make its Data Center Operating System easier to use for software developers, Mesosphere launched a software development kit and developer program.

Based on the open source cluster management system Apache Mesos, Mesosphere’s DCOS enables developers to write applications that can run on distributed data center infrastructure without being distributed-systems experts. It mimics the way giants like Google or Twitter use their infrastructure.

“Because it’s built with Apache Mesos at its core, the DCOS takes care of common distributed computing headaches such as job scheduling, high availability, resource isolation and networking — meaning developers don’t have to,” Derrick Harris, senior research analyst at Mesosphere, wrote in a blog post Wednesday announcing the SDK.

DCOS, the company claims, makes installing an application or a service in a data center or a cloud environment “as easily as you install an app on your laptop or smartphone.” Those are services like HDFS (Hadoop File System), the Kafka messaging system, or Cassandra, the popular open source NoSQL database.

Mesos was born at University of California, Berkeley, when Benjamin Hindman, one of its creators, was a PhD student there. Hindman now leads engineering efforts at Mesosphere, which he joined last year after four years at Twitter, where he oversaw implementation of Mesos in the social network’s data centers to reduce downtime.

In another example of a high-profile company using the open source cluster management system, Mesos orchestrates the underlying infrastructure for an internal Platform-as-a-Service Apple developers use to work on Siri, the natural-language interface for the iPhone.

Mesosphere rolled out DCOS through an early-access program in late 2014 and announced general availability only last month. According to Harris, the user base has been growing quickly, and the company hopes the release of the SDK and the developer program will accelerate adoption.

DCOS can be used to add distributed services in companies that already use it. Developers can also use it to write entire new applications, such as databases, file systems, stream processing, or monitoring tools, all of which will automatically have the ability to scale across clusters of servers.