I have been a long-time attendee of OpenStack Summit (since the days when it was split into a Developer Summit on Monday through Wednesday and a Business Summit and Conference on Thursday and Friday). Now the entire event is called the “OpenStack Summit” and developers, operators and business executives converge in one place to talk all things OpenStack. So far, Every six months the conference has grown – the most recent brought nearly 3,000 attendees together in Portland to talk about the future release of OpenStack, Havana, and to share their experiences with OpenStack to date. With this type of growth, it was only natural for the OpenStack Foundation to ensure there were multiple tracks so attendees can align themselves to the sessions and topics they care about most. Some of the tracks this year included:
- Business Value
- Case Studies
- Community Building
- Getting Started
- Operations Summit
- Project Updates
This is always a great way to put presenters and attendees in the same room to maximize the value of the conference. However, there is unofficial track that I consider to be among the most important: the “Hallway Track.” Loosely defined, the Hallway Track encompasses all of the conversations that occur in the hallway or other social gathering points throughout the conference. It is here that you get to hear the dirty details of what a particular group really thought about a presentation or idea. It’s here that the tough questions get asked, and mostly answered; questions that would have probably been better if asked during the actual presentation. It’s a safer venue, a smaller one, that is filled with trusted opinions. Large takeaways from presentations are commonly disseminated during the Hallway Track and those are the ideas that attendees take back to their respective companies (and likely include in their conference reports).
Key Discussion Topics
I’m not going to dive into the details of every Hallway Track session I attended, but I want to share some of the overarching themes that were discussed during these sessions:
- Essex adopters are running OpenStack clouds with production workloads today. There were two keynote sessions that highlighted users who got their first real taste of OpenStack during the Essex Summit: Comcast and HubSpot.
- Most talked about session. State of the Stack April 2013 presented by Randy Bias from CloudScaling was the most talked about session in Portland. It was raved about. It was an update on all the projects, their best parts, their not so great parts and some general thoughts on the current state of all OpenStack projects.
- Interoperability, oh my! Analyst sessions, panels, blog posts and snarky comments abound all week. A lot of the OpenStack community contingent still believes OpenStack interoperability will occur, but there were some very strong opinions floating around about OpenStack just being too large, having too many deviations and there being too many factors to truly bring the whole community to an interoperability consensus. One thing both camps agreed upon was the great potential of interoperability. Personally, I think it’s too early to make a judgment call, but after seeing so many products start to converge towards a more common codebase, it might not be a pipe dream.
- Whiteboards in design session rooms. This seems to come up every conference, but if a room is designated for a design session, it should include a whiteboard. Sure, mostly there is debate and healthy conversation with an etherpad, but sometimes a discussion calls for a whiteboard for architecture or process flow.
- If you’re not using SCM, you’re doing it wrong. The age of logging into servers or virtual instances is seemingly coming to an end. More and more companies talked about advanced automation and the replacement of instances rather than nursing them back to health.
- Federated Auth in Keystone. This is one item that is near and dear to Rackspace, as we have proudly been running OpenStack Grizzly code in production for weeks. But since we switched our cloud platform over to OpenStack (on Aug. 1, 2012) there were some key features that we ran differently than the community due to our legacy cloud customers. The biggest hurdle to fully adopting native OpenStack is authentication, as Rackspace is not running the community variant of Keystone, but a Rackspace-ified version. With federated authentication getting closer to reality, it will allow Rackspace to adopt Keystone fully into our platform and move that much closer to running native OpenStack code.
- There was not a Piston or a Nebula party. Let’s be honest here; when the agenda came out and both of these companies were missing from the evening events lineup, there was quite a stir. These companies know how to put on a great party and roughly 3,000 attendees were let down. Luckily, there were still a slew of great evening networking events hosted by Mirantis, PuppetLabs, HP, Red Hat and Rackspace; but it just wasn’t the same without Piston or Nebula putting their respective party spins on the OpenStack Conference.
- nova-conductor: One of the new services in OpenStack Nova, there was plenty of talk about what it does, why it’s needed and the best way to implement it. Yun Mao wrote a great post right before the Summit that answered a lot of those questions, and it was commonly referred to throughout the week. You can find his blog post on his personal website here: http://cloudystuffhappens.blogspot.com/2013/04/understanding-nova-conductor-in.html
- OpenStack Operations Guide. A key element of Jim Curry’s keynote presentation was that 2013 is the “year of the user” for OpenStack. Over the past two years we have all been part of a transition that started with a strong focus on developers and now adds a heavy dose of OpenStack users. While the development of OpenStack has never been stronger – there were 500-plus contributors to the Grizzly release; OpenStack has also experienced a tremendous spike in growth and adoption. Another signal that users are here and ready to play is the release of OpenStack Operations Guide. There are more options than ever to get OpenStack running in your data center, and knowing all of the components of OpenStack and how they interact has never been more important. Once you have a grasp of OpenStack, the next question that is “what is the best deployment model for me?” OpenStack Operations Guide answers that question. The book is absolutely free and can be found here. Even cooler, this entire guide was written in a five-day sprint before the conference.
OpenStack Products Abound
There was also a great deal of product talk at OpenStack Summit Portland, with many OpenStack players offering up new solutions. Check out NetworkWorld’s roundup of the hottest OpenStack products right now. Here’s a sampling of some of the products launched and discussed at the Summit:
- Rackspace announced the Rackspace Global Cloud Network: Rackspace revealed plans to expand its global cloud network through a program in which Rackspace will build and run interoperable public clouds for service providers around the world.
- RedHat announced RDO: RDO is a community-supported distribution of OpenStack that runs on Red Hat Enterprise Linux, Fedora and their derivatives. Red Hat hopes to do for OpenStack what it did with Linux.
- Project Savanna: Red Hat, Hortonworks and Mirantis will collaborate to contribute significantly to Project Savanna, Apache Hadoop on OpenStack for simpler and cost-effective transitions of big data workloads between public and private clouds.
- Nebula released Nebula One: Nebula finally came out of the darkness and released its long awaited product, Nebula One, a turnkey private cloud system that provides compute, network and storage services through a self-service interface and APIs, using industry-standard servers from vendors such as HP, IBM and Dell.
- CloudScaling announced Open Cloud System 2.5: CloudScaling launched Open Cloud System 2.5, which supports OpenStack Grizzly; 100 percent community OpenStack – no forks; virtual private cloud via Juniper Contrail technology with block storage snapshots; and runs on certified hardware from Dell, Cisco, Quanta, Juniper and Arista
- Mirantis FUEL: FUEL is the public open source release of Mirantis’ toolset for production deployments of OpenStack. It has been used in projects completed at NASA, PayPal, WebEx and others.
- Ceph is hot, Swift is not: A full replacement alternative for Swift, Ceph is an open source, distributed storage system that provides block and object storage fully integrated with OpenStack. You can find more information about Ceph on the Inktank blog about Ceph’s Cinder compatibility. That being said, SwiftStack put on another series of fantastic workshops that gave newcomers and grizzled veterans of the technology some hands-on time and theory around using Swift in production.
Overall, OpenStack Summit Portland was an excellent five-day look at the state of the project from nearly every perspective. Roughly 3,000 people showed up to shape, drive and learn more about OpenStack, and the energy and enthusiasm in the sessions, and the Hallway Track were infectious. One thing is for certain, OpenStack is here to stay. I can't wait to do it all again in six months.