The OpenStack community gathered in Hong Kong in the first week of November to define the roadmap for the upcoming Icehouse release cycle and reflect on Havana, the release that forms the basis of the upcoming Red Hat Enterprise Linux OpenStack Platform 4.0.
This was the first time the biannual OpenStack summit had been held outside of North America but was still the largest ever with over 3500 stackers in attendance. The OpenStack Foundation rose to the challenge, organizing yet another exemplary event. Here at Red Hat, we’ve spent some time gathering up a grab bag of our personal highlights for the week.
OpenStack Deployment and Management
Deployment was front and center with the Red Hat keynote setting the scene, taking a look at the emerging TripleO project.
Keith Basil, a Red Hat OpenStack product manager and redhatstack.com contributor, played the role of “shark food” in a lively roundtable on Deploying and Upgrading OpenStack. The roundtable faced off proponents of various deployment solutions to discuss both the present and future state of OpenStack deployment technologies.
In addition to the roundtable, many side conversations related to Deployment and Management were had by players in this space. The impact of TripleO/Tuskar was starting to sink in and prompted very passionate responses from those who perceive OpenStack Deployment as an official program destined to surplant their own internal and non-OpenStack community driven tools. Red Hat went through a similar phase with its own Foreman tool. The take away and consensus seemed to be that we’ll continue to develop and support what we have today until TripleO/Tuskar becomes more mature and stable.
We also provided a working demo on TripleO and Tuskar at the booth. This demo showed a rack of hardware supplied by Quanta being described and provisioned by TripleO and Tuskar. This was a phenomenal culmination of intense engineered work done by Red Hat and was well received at the Summit. It was great to be able to have conversations with community members and then have them at the booth for a live demonstration of the work done to date. There were many sessions on OpenStack “deployment” which reinforced both the validity and direction of the work being done around TripleO/Tuskar.
At a deeper level, traction was gained with the Ironic project. Ironic was spun out of Nova as an abstraction layer and specifically addresses the bare metal provisioning requirements within the scope of deployment. Red Hat expects to commit engineering resources to Ironic so that the lower levels of hardware integration are stabilized and feature complete enough to be meaningful to our ecosystem partners.
There is a tremendous amount of buzz related to OpenStack Deployment and we’re excited to have a substantial role in making the community vision a reality.
Software Defined Networking
Software-defined Networking (SDN) was also a hot topic at the Hong Kong summit. Alongside the OpenStack Networking (Neutron) design sessions, SDN was well represented in the general summit tracks as well.
Red Hat’s own Chris Wright presented OpenDaylight: An Open Source SDN for Your OpenStack Cloud alongside Kyle Mestery (Cisco), Anees Shaikh (IBM), and Stephan Baucke (Ericsson).
With the support of the Linux Foundation OpenDaylight is an open source project aiming to drive adoption of SDN by providing a robust and extensible SDN platform to facilitate industry innovation. In this presentation the goals, architecture and roadmap of OpenDaylight were presented. A vision for how OpenDaylight might integrate with OpenStack Neutron in the future to provide a powerful SDN-based networking solution for OpenStack Clouds was also provided. OpenDaylight was also on display in the Demo Theater.
Block and Object Storage
The Storage scene was very busy this Summit, Cinder design sessions were spread across two days. Everything we talked about was captured in the Icehouse Etherpad. The discussions were interesting, productive and covered various topics and problems pending to be solved, all the way from Taskflow and “More flexible scheduler policies” to “Changing the wheels on a moving bu.s” Cinder continuous deployment as well as several sessions around disaster recovery topics such as: Incremental Volume backup, Continuous volume replication, Backup as service and Stateless Snapshots
The OpenStack Performance and Benchmarking themes continued this summit, with several user sessions, covering a range of performance considerations important when sizing and deploying OpenStack configurations as well as testing best practices and Object Storage Real World Usage.
Summit panels covered Software defined Storage and Big Data that has seems to become a first class citizen in OpenStack.
The Savana project which aims to offer scale-out, on-demand Hadoop provisioning and management, received significant interest and airtime at the summit. The Savanna demo theater was well attended, as was the user summit presentation.
Metering and Orchestration
During the Havana release cycle the Metering (Ceilometer) and Orchestration (Heat) projects graduated from incubation, as a result they are now fully integrated OpenStack projects. Ceilometer collects usage and performance data from the various OpenStack services, while Heat facilitates orchestration of complex deployments on top of OpenStack clouds.
The changes to these two services as they matured throughout the incubation period, and the way they now combine to provide automatic scaling of resources in an OpenStack cloud, were captured in an excellent deep dive presentation “Ceilometer + Heat = Alarming”. A live demonstration of these facilities being used to deploy and automatically scale an OpenShift PaaS installation was also provided during the summit.
As always, the OpenStack Summit is all about participation & collaboration. We were pleased to be able to contribute and represent Red Hat at this event and look forward to the next Summit in Atlanta!