Kirk Klasson

Openstack: Here we go again….

OpenStack….monetizing enterprise computing won’t get easier but maybe other things will.

Several years ago I was having a cup of coffee with a business development exec from Jamcracker, back then an early instance of a rentable hosting environment. The proposition on the table was the incorporation of an open source operating system where the revenue would be extracted based on utilization; no customers, no revenue; some customers, some revenue but essentially no revenue until someone showed up. At the time, Jamcracker was trying to create a service delivery platform that could be affordable for both SMB and Enterprise class customers. Only thing was that try as we might, we couldn’t get the economics to go down market as there was still too much proprietary software in the stack. And bigger class customers had already achieved better economies of scale than this proposition could provide.

At the time, folks were aware that service orientation would eventually cause scale-out architectures to overcome scale-up architectures but a critical mass of scale-out based applications (think network assembled web-based apps) had yet to materialize. Virtual environments had only started to gain traction and the notion of a service-based, virtualized resource management environment (IaaS) was only beginning to emerge.

In fact, several firms believed that resource management was the last defensible proprietary bastion that remained in the conventional model of the enterprise software stack. And recent research would support this contention. The brightest spot when it comes to growth in enterprise software according to IDC is in Virtual Machine and Cloud System software, a trend that was by most measures sustainable, at least up until a couple of weeks ago.

Flash forward several years and the economics of rentable hosting environments began to work when suddenly most of the stack could be replaced by open source solutions. Only trouble was convincing SMB types, who happened to be end users, not SaaS suppliers, to move to what was then a nearly completely open sourced environment was still a daunting challenge.

Future Inevitabilities

Earlier this March, IBM, already an active project participant, publically announced its endorsement and support for OpenStack, a promising but still fledging alternative to proprietary virtualization and resource management environments. Reminiscent of its endorsement of Linux back in 2000, IBM had already cozied up to Red Hat’s KVM environment and instead of reading the writing on the wall decided to put it there themselves. Essentially, this is the strategic equivalent of beggaring your neighbor as IBM probably wouldn’t have gone this way if it wasn’t already convinced that Microsoft and VMware had more to lose than they did, and that the IBM – Red Hat alliance would benefit in the long run. Only thing is, an ebbing tide lowers all boats, and you wouldn’t do this unless you were comfortable that you wouldn’t end up foundered in the bargain, even though your competition may well end up on a sandbar.

There are, however, some differences between this endorsement and IBM’s endorsement of Linux back in 2000. For starters, Linux was a more seasoned body of code than OpenStack at the time IBM decided to endorse it. Linux was also a more cohesive and monolithic code base; OpenStack is made up of 7 primary components, which will likely introduce a greater degree of integration complexity to the project. And lastly, while the project has had early commercial supporters, the pedigree of critical portions of the code come from NASA, yep… that’s right, government code, which is why all Mars rovers eventually end up running in safe mode.

Linux didn’t really hit its commercial stride until about 3 years after IBM’s initial endorsement. Back then open source was relatively new and unchartered territory with backers still going through the courts on different IP and ownership issues. Today, the whole open source phenomenon is more mature and the OpenStack project has gained significant momentum very quickly due to a host of backers, not just IBM. However, adoption is likely to remain slow outside the commercial hosting space, unless of course you have some rocket scientist on staff who happens to be intimately familiar with critical portions of the code.

In the long run, we’re all dead…

Years ago, one of the pioneers of the open source movement was asked what his ambitions were for the company that he had founded. His answer was something like, and this is not a direct quote …I’d like to take a $50 billion industry and turn it into a $10 billion company. A noble goal since at the time there were only about 20 software companies that exceeded a billion dollars in annual revenue. Turns out, this never happened but the concept was valid and open source did prove to be a governor on future revenues for this particular segment of the enterprise software market. Which has also been largely the case when it comes to Linux, a late entrant into what was already a mature and saturated market, namely operating systems.

Introducing low cost substitutes to mature markets usually has the affect of diminishing the incumbents prospects, unless of course the substitute can actually create a new category or a new market that was previously not visible or viable. ( see The Ghost in the Machine and Requiem for a Business Model …January and November 2011) Open source has in fact gone from a mere low cost disruptor to the progenitor of new opportunities. With respect to open source operating systems, it wasn’t Linux, but its little cousin, Android, that set the smart phone market on fire.

So, will OpenStack present a similar opportunity?

Open source operating systems sparked a new opportunity when they jumped from the enterprise to the consumer world becoming the dominant player in hand-held computing and, it’s time we all faced it, the phone is the least smart part of a hand-held computing device. But it is highly unlikely that hand-held devices will need the equivalent of an OpenStack resource management environment.

However, it is possible that the consumer equivalent of a personal cloud might also emerge if it were inexpensive and easily managed. I would no sooner ship my personal data to a commercial cloud if I could have the equivalent functionality at my fingertips and in my living room. Granted, a virtualized resource management environment designed for the likes of Amazon or Rackspace is not something that your average consumer needs. But then again, Linux and Android may share the same lineage but not the same footprint.

So let’s see, a couple of Buffalo drives and some Raspberry Pi, a little open resource management and viola! … Personal clouds.

Graphic courtesy of SuZQ Art and Images

Sorry, the comment form is closed at this time.

Insights on Technology and Strategy