The cloud may provide a level of flexibility and scalability that traditional data infrastructure can never hope to match, but as a practical matter it still represents the same challenges as any newly deployed resource architecture. That is, enterprises are still confronting a monster data migration effort.
To be sure, all of the leading cloud providers have ample means of moving data into their platforms and then between the various SaaS, PaaS, IaaS and whatever-aaS environments are made available. Ideally, however, the full scope of cloud resources should be at your disposal, not just those of a single provider.
Still, to gain that kind of flexibility, the cloud would have to embrace a level of openness that has eluded both the physical and virtual layers of enterprise infrastructure. From the looks of things, however, it appears a truly universal open cloud environment is still a pipe dream given the ongoing turmoil surrounding the leading effort, OpenStack.
Although the format still has the backing of industry leaders like Rackspace and HP, it took a hit this week when Citrix decided to release its CloudStack IaaS platform on the Apache Foundation, a move that is expected to speed up third-party code contributions due to broad support of the Apache GPL 2.0 license, as well as greater API compatibility with proprietary cloud platforms like EC2.
One of the biggest complaints against OpenStack is that it's simply not ready for prime time. The next release of the platform, dubbed Essex and due out later this month, is said to be the first commercially viable version, although there is growing criticism that it will still lack critical functionality, which will only have to be addressed in next fall's Folsom release. Expect to hear a lot of discussion along these lines at the OpenStack Design Summit later this month.
Meanwhile, numerous other open cloud initiatives are following their own development paths, each laying claim to be the premiere solution for widescale distribution. VMware, for one, touts its broad reach into data center infrastructure as the strongest indicator of universal interoperability, while providers like Red Hat tout the need for open source code as a basic prerequisite for an open cloud.
It seems, though, that with the cloud, opening up just the source code will not be enough. APIs play a significant role in cloud functionality, as do data formats themselves. If the goal is to shift workloads across multiple platforms and infrastructures to better match operational needs as opposed to technical ones, then enterprises would do well to build broad commonality across the entire data stack.