Embracing OpenStack

In the world of clouds, there are really only three giants: AWS, Google and Microsoft Azure. By default, OpenStack is the poor man's cloud software, with a much lower cost than writing a proprietary stack. That's not to say that OpenStack is a poor solution. Quite the opposite! While it's still evolving and growing in scope, the current code release is well featured and the basis for a production public cloud system.

Established system vendors are now seriously contemplating a future where much of their business derives from delivery of cloud services. Offering a mix of public cloud and hosted private clouds together with on-premise private clouds, the traditional players will use a software and services model to replace the hardware sales being lost to commodity-class manufacturers from China. As a higher margin business with a repeating revenue stream and high growth expectations, the cloud services sector looks like a safe haven in the coming storm of change.

As a result, OpenStack is turning a corner. None of the major systems vendors are in a position to generate their own stacks, having come late to the realization that they couldn't copy Canute and hold back the inevitable. They face Hobson's choice, with OpenStack the only non-proprietary orchestration and management system with the scope needed for a full public/private cloud deployment.

Consequently, OpenStack is getting a strong boost in support from the likes of HP and Dell, and looks to be the solution of choice for a lot of clouds. With support from software companies such as Red Hat, VMware and with Cisco on board for networking, OpenStack is in an increasingly strong position in the cloud market.

Of course, as in most things they do, the established players are adding bells and whistles of their own. OpenStack already has "flavors" with a range of distributions that add other tools and proprietary code to the common core. In this respect, OpenStack is a bit like Linux, and, like Linux, the number of distributions will shrink back in a few years' time.

Traditional system vendors are torn between keeping their proprietary code tightly held or moving some of it to the public open-source domain as part of the common OpenStack offering. They are torn between strengthening OpenStack in meaningful ways, and holding back vendor lock-in hooks. This is a balancing act between giving important features to the common pool versus holding them tight while risking triggering customers' fear of lock-in.

Meanwhile, Microsoft is positioning Azure as a software product to create a hybrid cloud environment, sort of as the "Windows" of the cloud. The software giant wants to hold onto its server base and expand into the cloud.

Azure is a mature product and is keeping up with advances such as containers, so it can presents a strong challenge to OpenStack on the technical front. On the financial front, it's less clear, since there has to be a licensing fee involved with Azure. The cost, however, is offset since Microsoft handles most of the building of infrastructure and tools around Azure.

Azure has achieved some traction, notably with IBM. That will interest some of the players who are still trying to keep their options open, perhaps the major telcos. The choice of stack direction may well hinge on the issue of commitment to a single vendor (Microsoft) at a time when the cloud is still finding itself and is evolving rapidly.

For private/hybrid cloud administrators, Azure might mean an undifferentiated solution with the key software provider also being a major CSP competitor, which limits Microsoft's position in the battle for market share with OpenStack that looms in 2015/2016.

Choices in cloud stacks mirror the market dynamics with a conservative "Windows" approach versus a rebel "Linux" cadre. Since it doesn't seem likely at this time that AWS and Google will offer a software-only solution for private and hybrid clouds, OpenStack and Azure will split the cloud software market between them, but the question of whether OpenStack's open nature and inherent modularity and extensibility are a winning combination will take some time to decide.

Clearly, a lot of the established system vendors are betting that OpenStack will win.