Why Mirantis is Excited about OpenStack Drivers
Today we officially launched the DriverLog project and an associated dashboard, showcasing the detailed list of available OpenStack vendor drivers. We worked with Project Technical Leads and the OpenStack foundation on this effort and are glad to see that many vendors have already started contributing and maintaining their driver information.
Meh… drivers… boring. “Why bother?” you may say. But bear with me, and I’ll make it a bit more entertaining for you.
First, let’s step back and look at the biggest success story in the history of open source – Linux.
The late 70s and early 80s were the days when computing started moving from defense and niche sectors into the mainstream enterprise. At that point, big guys like IBM were selling mainframes, which tightly integrated hardware, operating system and even software into a single package. The market started maturing, x86 architecture went mainstream and with it came a need for a new operating system fabric: one that would enable end users to quickly aggregate and take advantage of all the innovations in a disaggregated and fast moving x86 world. That fabric needed to be more flexible with respect to the underlying components as compared to the hard-bundled mainframe offerings. It needed to run on hardware sold by various vendors and assembled from parts made by discreet and focused innovators in their respective niche areas (CPUs, memory, network cards, storage, etc.)
Microsoft was born. It became very big, very fast. So big, in fact, that many vendors got scared and a window of opportunity emerged for an open, competing standard that could to help deflate the Microsoft monopoly. Linux emerged as that open standard. The reason people wanted Linux was partly because it was open, but, more importantly, it offered that same flexibility for the underlying hardware that Microsoft offered in Windows. Linux became the open source poster child of the OS/Hardware disaggregation stage in the computing market.
What we saw with mainframes, Microsoft and eventually Linux were three examples of the same phenomenon. To seed a new, disruptive market you always need to start with a vertically integrated, tightly packaged and easily understood offering that delivers immediate value. Examples:
Mainframes – to seed enterprise computing
iPhone – to seed mobile
Amazon EC2 – to seed infrastructure cloud
Once the market is seeded and starts expanding, a single vendor that caused the initial disruption can no longer keep up with the innovation of the broader market and keep its monopoly. Expanding markets then go through a stage of “disaggregation.”
What we are seeing today is the new cycle of disaggregation happening in the computing market. The industry is moving from client/server to fully distributed (or in cool marketing speak, from Operating Systems to Software Defined Datacenters). With this disaggregation, we are seeing a lot of innovation happening under the “software defined” label – software defined storage, software defined networking, etc. Unlike with Unix and mainframes where it was OS/hardware disaggregation, today’s disaggregation is mostly happening in the layer above the hardware.
As the market goes through this, OpenStack has emerged as the next generation operating system fabric that glues together the innovations of the distributed, software-defined world.
According to the recent OpenStack user survey, flexibility of underlying components was among the top business drivers for adopting OpenStack. Double-clicking on this, I’d argue that all other factors listed in the top 5 are generic to cloud or open source, making flexibility of the underlying components the most important differentiator of OpenStack platform today.
When you look at OpenStack through that prism, it becomes evident that the disruptive nature of this platform is largely defined by the number and quality of integrations with underlying vendor solutions / innovations. Drivers are the point of entry.
And that is a very big reason to be excited about OpenStack drivers.