Description of icon when needed 6 Min Read

Another OpenStack Summit is in the books! Last week, I attended my 13th OpenStack Summit in Boston, Mass. Part of what makes OpenStack Summit such a great event is the mix of activities and people. While I saw a lot of familiar faces and spent a lot of time catching up with old friends, I also met a lot of new folks—in fact, about half of the attendees at each OpenStack Summit are there for the first time

While I spent some time catching up on projects I’ve been working with for some time, I also spent some time with relative newcomers to the scene and adjacent communities. In fact, this spring OpenStack Summit hosted a number of “Open Source Days” for the first time. Open Source Days brought adjacent communities to the OpenStack Summit so attendees could get the best of all worlds. The roster included Kubernetes, Open vSwitch, OPNFV, CloudFoundry, and more. Specific use cases were on display too, with a lot of sessions and hallway conversations focussing on NFV and container workloads atop OpenStack.


If that wasn’t enough open source, adjacent open source technologies also made it to the main keynote stage the second edition of the Interop Challenge. I joined a large group on stage to do a live demonstration of a cloud-native workload deployed without modification on 15 different OpenStack clouds. This time around, we enlisted a little help from our friends at CoreOS and CockroachDB. We deployed Kubernetes on CoreOS guest instances, and then spun up a containerized CockroachDB cluster on top. While that in itself was a great testament to how much more portable workloads on top of OpenStack have become and how diverse the ecosystem is, we also spent some time this year to demonstrate not just repeated copies of the same workload, but a single app running across clouds. VMware joined a group of others to connect our CockroachDB cluster into a single large, geo-distributed database cluster spanning several clouds. As we see hybrid cloud or multi-cloud use cases go from thought experiment to production deployment pattern, this demo was a timely reminder of how fast the cloud ecosystem is evolving—and how fast OpenStack is moving to keep up. OpenStack is now approaching it’s seventh birthday, while Kubernetes is just shy of two years on from its 1.0 release and CockroachDB launched it’s 1.0 release on May 10th—the day after our on-stage demo!

While a lot of energy went into use cases in which OpenStack is rapidly growing and adjacent open source technologies that OpenStack users are increasingly adding to their portfolios, plenty of time was also spent on traditional OpenStack projects. Boston was the first OpenStack Summit under the new Summit format. Following last year’s OpenStack Summit in Barcelona, the community split the Design Summit portion of the conference out into a separate event called the Project Team Gathering (the next of which will be held in Denver later this year) so project developers could spend more time focussing on development.

The split also freed up some time at the Summit for developers to spend more time in “fishbowl” sessions with users—a new series of events within the Summit called “The Forum”. The Forum sessions I attended were mostly spent gathering feedback from operators and users about what to work on next to make sure OpenStack remains a good fit for production use. Generally, the feedback seemed to focus on a few key themes: emerging use cases (such as NFV and containers) and continued hardening of widely-deployed projects (like Nova, Cinder, and Neutron).

One particular topic that generated a lot of interest was a new addition to one of the oldest OpenStack projects: Nova. In the new Ocata release of OpenStack, all Nova deployments use Cells v2, a reimagined version of the experimental Nova Cells feature that was introduced years ago as a way to help Nova scale to larger deployments. Starting from Ocata, all Nova deployments will be a “cell of one”, which support for multiple cells landing in future releases. A fair amount of time was also spent on the Nova placement API, which was initially introduced in the Newton release and is increasingly important to scaling up deployments and alleviating expensive rescheduling events from the Nova scheduler.

All in all, Boston was another great OpenStack event: a steady hum of activity mixed with a few new twists.  We’re looking forward to the PTG in September, and the next OpenStack Summit in Sydney in November—see you there!

Don’t miss Misbah Mahmoodi’s deep-dive into how telcos can benefit from open source platforms like OpenStack. Stay tuned to the Open Source Blog for more.