Four years and change after its debut, OpenStack has reached its 10th biannual release, code-named Juno. With this release comes not only the usual flood of new features and bug fixes, but what the OpenStack Foundation hopes is proof of how it's refining OpenStack to appeal to a broader audience.
One way this has already taken shape is to build support into OpenStack for more of the applications businesses are running at scale. Earlier versions of OpenStack added the Trove component to perform provisioning and management functions for common databases (MySQL, MongoDB, PostgreSQL, Couchbase). Juno expands on the functionality for databases, but also contains a new provisioning service, code-named Sahara, for Hadoop and its Spark in-memory processing system.
Upgrades, long a pain point for OpenStack users, were simplified in the previous iteration of OpenStack, but in a limited way. Juno has broader support for rolling upgrades so that running applications aren't impacted as heavily.
Juno also introduces storage policies, which allow OpenStack admins to set a balance of their choosing between reliability of storage (more copies made) or cost (fewer disks used). As Mark Collier, COO of the OpenStack Foundation, explained in a phone call, this feature allows an application to target different classes of storage based on its needs.
But many major new features in Juno appear to have been driven heavily by telecommunications companies and other industry verticals that are becoming the heaviest adopters of OpenStack -- and, consequently, exerting the biggest influence on its development. Network function virtualization, for example, has become a popular feature with OpenStack, mostly for telcos seeking an alternative to older fixed-function hardware. In fact, it's so popular that an entire working group, with input from OpenStack-adopters like AT&T and Ericsson, has been formed to further shape it.
Collier believes all users of OpenStack, not only telcos, benefit from these features. What he describes as "one class of users pushing the envelope" results in a better product for everyone, because it teases out more bugs and raises the bar overall. "Having a high bar for quality, performance, and real-time interaction from the hypervisor up to the end user is beneficial to all users," he said.
He believes the real killer app for OpenStack will be its speed -- not only in processing, but in providing resources on demand for an organization. This partly squares with the feedback most enterprises give for why they use OpenStack. But though some companies adopt OpenStack as a cost-cutting measure, it's tougher to determine if the value is derived from OpenStack providing additional business agility or to use its a substitute for commercial products like VMware.