Last week I was thrilled to participate in the Linux Foundation 2010 Collaboration Summit panel "Does Open Source Mean Open Cloud?" The panel was moderated by John Mark Walker director of community at Splunk, and I was joined by James Urquhart, senior market strategist for cloud computing at Cisco and author of the "Wisdom of Clouds" blog at CNET; David Lutterkort, RedHat architect on the DeltaCloud project; and Doug Tidwell, senior software engineer at IBM working in their standards group. Great panel with a variety of perspectives on one of the hottest topics in cloud computing today.
During our conversation, it quickly became clear that the presence of open source technologies in cloud computing does not guarantee the freedoms, interoperability, portability and collaboration that have been the goals of the open source movement. As Urquhart pointed out, open source technology is a critical and ubiquitous part of the cloud stack, and major providers such as Amazon are actively making contributions to the community. Ultimately though, there is a difference between open source and BEING open- for example, the fact that large providers may be using open source does not guarantee the production or acceptance of upstream contributions to code.
The entire panel agreed that this really comes down to data openness- as Tidwell pointed out, it's not just about infrastructure- it's also about achieving openness further up the cloud stack at the level of services and applications in ways that give users freedom, ownership, access and portability of their data. The cloud model- including the addition of latency and the ability to change open source software and host it in your own own cloud- imposes new constraints on data openness and changes the landscape of code contribution.
So if open source doesn't mean open cloud, what can, will and should be done to ensure it? Walker noted that this will involve making a decision that open source and the principles behind it are important, and ensuring that there is incentive for companies to create and participate in an open cloud. Many on the panel believe that the creation of an open cloud will occur when small providers come together to form a larger ecosystem founded on principles of openness- Lutterkort made a strong point that this can be achieved in the private cloud space as well. There's already signs of such community gathering focusing on open cloud, with projects and efforts like jclouds, libcloud, the Open Cloud Consortium and DeltaCloud (see my blog on Open Source Cloud Projects). I think that open APIs, more simplified standards of data ownership, and increasing focus on the application and services layer will be a major part of these movements.
I invite you to check out the panel and let me know your thoughts!