OpenStack's Stefano Maffulli is on the opening keynote panel at CloudOpen Wednesday morning August 29. He will be joined by Mark Hinkle from Citrix, Greg DeKoenigsberg from Eucalyptus and John Mark Walker from Red Hat on a panel titled "What is the Open Cloud: Stacks, Platforms and Open Source." Maffulli was generous to take time out of his busy schedule as Community Manager at OpenStack to share his ideas on what is an open cloud and what's new with his project.
You are participating in a keynote panel about cloud stacks and platforms that will explore what is the open cloud. Can you give us a teaser of the position you plan to take as one of the keynote panelists?
Maffulli: 'Open Cloud' is a new term. Like 'open source' at the time it needs a clear definition or we risk it becoming an empty one. OpenStack defined very early in the process three basic tenets for developing software and its community:
- Open Design. We are committed to an open design process. Every six months the development community holds a design summit to gather requirements and write specifications for upcoming release. The summits are open to the public and include users, developers, and upstream projects. We gather requirements and produce an approved roadmap used to guide development for the next six months.
- Open Development. We maintain a publicly available source code repository and patch review system through the entire development process: no private branches, no big dumps of code before a release. This makes participation simpler and allows users to follow the development process and participate in QA at an early stage.
- Open Community. One of our core goals is to produce a healthy, vibrant developer and user community. Anybody can join, and most decisions are made using a lazy consensus model. All processes are documented, open and transparent.
The OpenStack community makes the following promises:
- The community will be involved in the design process. You can help make this software meet your needs.
- The community will have representation on the technical board, which has the ability to override decisions by the project lead.
- This will always be truly free software. We will never purposefully limit the functionality or scalability of the software to try and sell you an "enterprise" version.
- All project meetings will be held in public IRC channels and recorded.
Developing OpenStack respecting these principles is one piece of the open cloud. To complete the picture you need to have also other things like a pluggable architecture and good API and documentation to help users adopt your product. The end goal for the "open cloud" is interoperability and no vendor lock-in, which will be the payoff for these open development efforts and industry collaboration.
OpenStack has received a lot of attention and support this year. Can you give us a brief "State of OpenStack?" Give us an update on how the project is going.
Maffulli: This has been a very exciting year for OpenStack and the momentum continues to grow. We have a very strong community that has put an incredible amount of work into the technology, community building and formation of the OpenStack Foundation, which is expected to be fully operational in the next few months.
To provide a few statistics, we've had 200,000+ software downloads from the central repository, not counting the growing number of distributions. We're very excited to have the firm backing of major Linux Distributions: Debian, Fedora, Red Hat, SUSE and Ubuntu, in addition to emerging players like Piston and StackOps.
We had more than 200 individuals from 55 different companies contribute to our latest software release, called Essex, during a six-month development cycle. Our next release, Folsom, is expected at the end of September and will feature two new core projects focused on networking and block storage.
There are more than 100 known OpenStack deployments at service providers, research institutions and enterprises across the globe. We now have more than 180 participating organizations, including newcomers this year AT&T, Brocade, IBM and Red Hat.
In less than two years, the OpenStack Design Summit has grown from 75 people to 1,000+ attendees. More than 20 different countries were represented at April 2012 Summit, and user groups and conferences have been organized across the globe. More than 1,600 attendees are expected at an upcoming conference in Beijing and Shanghai, August 14-15.
Which OpenStack-specific sessions at CloudOpen would you recommend to attendees?
Maffulli: I'm trying to select the schedule for myself. There are so many interesting talks and people to meet. I'll definitely go listen to my colleague Tomas Muraus with his talk about how to avoid vendor lock-in using libcloud.
What the APIs mean for the 'Open Cloud' is still something that needs to be defined. Some of OpenStack supporters believe that truly open APIs can only derive from truly open development as defined within the OpenStack project (that is, Amazon APIs cannot be considered open). With Google entering the space the concept of API for the Open Cloud gets even more complicated.
The GNU project and Linux had a much easier life. POSIX standards were already established; C library and toolkits were documented. All Stallman and Torvalds had to do was to re-implement and improve on something already existing. It wasn't easy, but they had a path to follow. With the cloud, open source projects like OpenStack are on the frontier of innovation, exploring totally new roads.
Where do you expect the OpenStack community to be a year from now? What are the goals and/or forecast for the project?
Maffulli: We're in the process of establishing the OpenStack Foundation and have already attracted more than 2,000 individual members during the first 10 days of the sign-up period. There are moor than 40 people nominated as individual members of the Board of Directors. I know OpenStack has a large and diverse following, but I still get surprised when the numbers come in.
More than a forecast though, it's a hope: I hope to see more products based on OpenStack running on ARM-based chips. I have a soft spot for small, efficient, elegant designs and ARM-based servers are just that. It'll be great to see innovation happening around hardware, too.
We're also continuing to mature our development processes. Creating a strong feedback loop between the user and development communities is a topic that I've seen pick up interest recently and that I look forward to seeing progress next year.