OpenStack is building up services fast, including some aspects of platform-as-a-service, in its new Havana release. That increases its appeal as the basis for private cloud in the enterprise.
There's been Essex, Folsom and Grizzly, and now there's Havana. The next biannual release of OpenStack makes several major additions in workload orchestration and service delivery. All in all, 392 features have been added.
But some observers think OpenStack is evolving rapidly into something much more than just new features added to existing infrastructure-as-a-service (IaaS). They believe OpenStack is beginning to track Amazon Web Services and add platform-as-a-service (PaaS) features for rapid software development too. If so, OpenStack may start to command a broader body of interest as the basis for private cloud in the enterprise.
That may prove to be a topic at the next OpenStack Summit, to take place in Hong Kong Nov. 5-8. Open source is popular in China -- no need to send more yuen to Microsoft -- and the summit organizers hope the Chinese location will increase OpenStack's profile there.
Of the IaaS features, a total of 20,000 patches have been added to upgrade previously basic services in OpenStack, such as metering for user chargeback or showback. Metering in previous releases "was limited," conceded John Igoe, VP of private cloud at Rackspace. It's been integrated into operations on a more thorough and supported level, said John Igoe, VP of private cloud at Rackspace, allowing managers to view more details around each individual user's bill. Rackspace is an OpenStack contributor and host to one of the largest implementations of the open source software, Rackspace Cloud.
Likewise, said Igoe, Havana features the Orchestration service, where virtual systems may be equipped with the compute, storage and network resources a user chooses from a template, then deployed onto OpenStack infrastructure.
Havana includes a driver for Docker, an open source code system that allows an application with multiple dependencies, such as an application server, database or Web server, to be packaged together in a single container and moved around, much like a large file. The container gives the application an additional measure of security, said Igoe, limiting the application to access only a file system packaged with it and running its processes with an added degree of isolation.
Havana includes more effective implementation of its quality of service capability. QoS in the cloud consists of measures that provide needed I/O or network bandwidth to the resources that need them. Havana makes QoS guarantees available to a variety of block storage services, provided they support OpenStack's approach to QoS.
In this release, OpenStack is showing the vigor of a young and growing open source project, as both Apache and Linux once did. It's not exactly the dead duck described by Simon Wardly in "The Trouble With OpenStack." On the contrary, it boasts code contributions from 910 developers, "up 60% from Grizzly," wrote cloud architect and CEO Randy Bias in his blog about Havana at Cloudscaling.
Buried in the details of the contributor list are new sources of contributions over the Grizzly and Folsom predecessor releases, including Canonical, Dreamhost, eNovance, HP, IBM, Intel, the OpenStack Foundation and Red Hat.
With code pouring into the project, Bias called out the work of OpenStack's own testing and continuous integration teams, led by Monty Taylor and Thierry Carrez, who set up the continuous integration melding new code into a core system and Jenkins testing processes. OpenStack needs to be thoroughly tested as it constantly assimilates new code, and it is, Bias asserted. "Fact: they're spinning up 700+ test clouds every day," he wrote.
Some observers cite both the development process now underway and the incorporation of other pieces of open source code, such as Heat, Trove and Marconi, as paving the way toward PaaS on a track similar to Amazon's. Amazon started out as infrastructure only, then started building out one development service after another for the infrastructure, such as SimpleDB and Elastic MapReduce. OpenStack is doing the same thing, said Alex Freedland, CEO at OpenStack consulting firm Mirantis, when it puts Heat application orchestration and Trove, OpenStack database as a service, on top of its infrastructure.
As an example of project vigor, 63 engineers from 20 different companies poured 90,000 lines of code into Heat for the Havana release. In the previous release, 97% of Heat's code came from a single source, Red Hat. With Trove, two companies -- Rackspace and HP -- accounted for all the code in the previous release; this time, they were joined by Mirantis, Red Hat, eBay and Suse.
"Clearly, OpenStack is moving up the stack into platform-level services, and the community is pushing it there," wrote Freedland. As it does so, it will become a more attractive platform on which to build enterprise private clouds, he predicted.
We welcome your comments on this topic on our social media channels, or [contact us directly] with questions about the site.