Too Many Cooks in the Open Cloud Kitchen?


These are heady days for those who've been calling for open-source platforms in the enterprise for the past few decades. Now that the cloud is poised to remake data infrastructure as we know it, open-source has emerged as not just the best way to foster interoperability, but the only way.

Of course, this has led to an embarrassment of riches, so to speak, in that the rising number of open-source cloud solutions is threatening to undermine the very intent of open-source: broad compatibility among disparate data platforms.

According to RightScale, nearly two-thirds of enterprises looking to build private clouds have decided to go with open-source solutions, with more than 40 percent opting for a strictly open-source environment. The question facing most enterprises these days is whether the various open-source consortiums will find it in their hearts to cooperate with each other, or are we heading for a fight to the death akin to the proprietary battles we saw during the 1980s and 1990s?

If reports coming from recent cloud conferences are any indication, however, there isn't much love lost between leading open-source proponents like OpenStack, CloudStack and Eucalyptus. Earlier this year, Citrix jumped ship from the OpenStack project in favor of CloudStack for its Apache platform, resuting in many hurt feelings. And as competition heats up to gain favor among leading cloud providers like Amazon and Google, tensions are understandably high as to who has gained the most industry support and whose APIs provide the broadest set of features and the most compelling user experience.

For the enterprise, assessing the relative merits of the various solutions will likely be one of the top challenges in the coming decade. OpenStack, for example, offers massive scalability and has drawn support from top names like Dell, Cisco and IBM. Meanwhile, Eucalyptus is said to have an edge when it comes to hybrid cloud deployments by virtue of its close relationship with Amazon Web Services. At the same time, there are numerous vendor-driven solutions like Red Hat's OpenShift, VMware's Cloud Foundry and the above-mentioned Citrix Cloudstack aimed at more specialized PaaS and IaaS functionality.

Part of the problem in nailing down an open cloud platform is that the virtual layer that it rests on is still largely proprietary, according to IBM's Robert LeBlanc. With various cloud services relying on multiple virtual machines, data exchange can be hit with any number of roadblocks, particularly now that VMware is out with its own open platform: Serengeti. LeBlanc says IBM is turning more toward the KVM format used by Linux because it provides smoother integration into open cloud platforms like OpenStack, which has drawn support from Linux vendors like Red Hat.

At the moment, it seems the cloud has plenty of room for growth as enterprises become desperate for ways to expand their data footprints without emptying their bank accounts. Ultimately, however, the open-source issue must be resolved, either by industry cooperation or survival of the fittest. Neither proprietary systems nor dueling open environments will allow the cloud to live up to its potential as the game-changing, dynamic data architecture of the future.