In April, the OpenStack Foundation released Icehouse, the ninth release of the OpenStack cloud platform. The general consensus seems to be that while Icehouse has some compelling new features, the focus of this version of OpenStack is on enterprise-grade readiness. Vendors like Rackspace and users with strong engineering talent have been very successful in creating enterprise-grade clouds using earlier OpenStack releases, but there has been a concerted focus by the community to stabilize the code and to enhance the operational capabilities of the platform. As a result, more vendors and partners in the ecosystem are now able to deliver OpenStack-powered products and services that can satisfy the stringent infrastructure requirements of enterprise shops.
Let’s take a look at what makes Icehouse enterprise-grade:
Improvements and Features of Note
Given the large number of new features and bug fixes in Icehouse, I won’t try to do a comprehensive review. Instead, I will highlight a few key updates and talk about their importance in creating an enterprise-grade cloud platform.
- Federated identity management: Federated authentication is now supported, allowing users to access private and public OpenStack clouds with the same credentials. This is very important for enterprises that will run multiple instances of OpenStack clouds but want to simply identity management across those clouds. It’s also another step in enabling federated management of hybrid clouds, built using OpenStack-powered public and private clouds; this federated management is key for enterprises that are required to store data in their own data centers but wish to leverage the public cloud for certain use cases and workloads.
- Enhanced platform stability: Past releases have introduced major new projects such as Neutron, which added valuable new services to OpenStack but also required major code changes in the core projects. In Icehouse, the only new integrated project is Trove (which I’ll discuss in more detail later), which adds new functionality and does not require major changes to core projects such as Nova or Keystone. Instead, there have been a number of code updates based on feedback from current deployments that have increased platform reliability and make OpenStack Icehouse an enterprise-grade release.
- Live upgrade of Nova Compute: Icehouse introduces the ability to upgrade OpenStack compute nodes in a rolling fashion so downtime is not required to upgrade compute nodes. In Icehouse, the controller nodes can be upgraded while the compute nodes are still running on Havana code. The compute nodes can then each be upgraded to the Icehouse code while cloud instances continue running on the compute nodes. This is a major functionality enhancement for enterprises that must reduce the amount of potential downtime for their infrastructure and expect live upgrades to be table stakes for any enterprise-grade technology.
- Improved user interface and experience for Horizon: The Horizon dashboard/web portal has been enhanced to provide admins and users an improved UI experience. While the expectation is that most users will interact with and manage their OpenStack cloud using APIs, there are cases where enterprise admins and users will use the dashboard to monitor and manage their OpenStack clouds.
- Project Trove (DBaaS) being integrated into OpenStack: Trove enables relational and non-relational databases to be created and configured on OpenStack, as a service and with minimal manual configuration, thereby simplifying typically complex database administrative processes. Databases are critical applications in the enterprise and Trove begins to give users the ability to consume databases as cloud services, making users more agile and more efficient.
Rackspace and Icehouse
Rackspace runs the world’s largest OpenStack-powered public cloud and also operates a number of OpenStack-powered private clouds on behalf of enterprise customers. As such, there is always strong interest in how we consume the latest OpenStack releases and what we are doing to operationalize the platform. Users and other vendors alike rightfully believe that Rackspace’s experience with OpenStack is a bellwether in judging the enterprise-readiness of the platform.
The Rackspace Public Cloud is currently running code that is based on an Icehouse feature freeze release candidate. That means that Rackspace effectively has all of the new capabilities of Icehouse running in our public cloud (for Nova, Glance and Swift). We should be running the actual Icehouse release code some time in the second quarter as we continuously update our cloud. We recently released new updates to the current Grizzly- and Havana-based Rackspace Private Cloud (RPC) product and service; Rackspace is now working on an Icehouse-based version of RPC and we plan to deliver an enterprise-grade version to the market in a timely fashion.
One of the major focuses in making OpenStack more enterprise-ready, along with the Icehouse code, is operational capabilities. These capabilities are not only provided through new features, such as live upgrades, but also by codifying and automating recommended practices for operating an OpenStack cloud. Rackspace is committed to sharing these practices with the OpenStack community and helping to increase the OpenStack community’s collective operational knowledge and excellence. Stay tuned for more details. Meanwhile, please attend the Rackspace keynote at OpenStack Summit Atlanta next week during which Troy Toman, Cloud Architect and Foundation member, will talk more about what it will take to continue the advancement of the OpenStack cloud platform.