There’s enough going on in the OpenStack ecosystem that you can pretty much find a comfortable niche drilling down on anything from hypervisor compatibility to driver support to who’s in the foundation. But as you find yourself working on IaaS, whether from a deployment automation perspective or just building out your cloud, it’s useful to take a step back and see where things are headed.
Two posts by members of the OpenStack Board in the last week or two highlight some key issues and opportunities. First, an interview with Mirantis co-founder Boris Renski at the Dell TechCenter blog by Dell Evangelist Rafael Knuth from Berlin, takes a look at the state of progress for OpenStack, with an eye on the way forward.
Rafael: What are the key accomplishments in the OpenStack project so far?
Boris: From the tactical standpoint, people oftentimes criticize OpenStack for the lack of certain features … unlike in vendor-centric projects … but I think OpenStack’s openness is its strength, which will become evident as the project matures. OpenStack is not an out of the box product, but rather a fabric aimed to ultimately glue together in an eloquent way a very diverse set of application infrastructure components. If you want NetApp storage …you can potentially do it with OpenStack. If you want Nicira networking … OpenStack will work. The whole idea behind the OpenStack design is a consistent set of APIs on the front end, where tenants can interface with. Effectively anyone can write drivers to plug their particular hardware component in the backend and make it work with OpenStack.
Rafael: Are there any gaps or missing links in OpenStack?
Boris: OpenStack is a new project, but I don’t think there are any clear gaps. Some functionalities are missing compared to Amazon or VMware solutions and the community needs to put them in place. But at Mirantis we always looked at OpenStack long term. OpenStack is the only project that is doing open cloud infrastructure right and that is the winning strategy long term. The features and functions will be there over time.
My key observation is that the right time to take the long view of OpenStack is now.
A second interesting discussion about the state of play for OpenStack is in video, courtesy of the Venturebeat CloudBeat 2012 conference. In this discussion, OpenStack Foundation Board Members Lew Tucker of Cisco (he’s vice-chair of the board) and Randy Bias of Cloudscaling (his blog a is a must-read in the world of OpenStack) look at the implications of OpenStack for compute applications that are already out there:
What’s most compelling about this conversation is how Randy contrasts OpenStack with legacy architectures that once seemed irreplaceable — in particular, mainframes. The fact that mainframe emulation can still run mainframe software does not mean that the mainframe going forward is the platform of choice.
The clear implication is that while monolithic big-box-failover platforms for applications can be virtualized, it does not mean that the future is about monolithic-big-footprint VMs running those same applications as-is.