If you think Converged Infrastructure & Fabrics are niche, guess again

A few weeks ago, I Tweeted about an analyst conversation where it was looking like the market for Fabric Computing / Unified Computing would be growing rapidly in the foreseeable future.

Another analyst friend of mine quickly commented back – sarcastically – that the market was sure to be in the billions of dollars.

I was feeling a little unsure about this market until a few weeks later when I was shown a technology report from Thomas Weisel Partners. Although the market definition for converged infrastructure (also known as Unified Computing) was still forming, TWP felt that sales of Converged Infrastructure solutions could rise as high as $15 billion by the end of 2014.  Billion with a “b”?  Right-on…

Then there is a report by Gartner Research on fabric-based computing… which estimated that by the end of 2012, roughly 30% of the world’s top 2000 companies would have some form of fabric-based computing architecture. (Under the heading of “fabric” falls Unified Computing as well as Converged Infrastructure).

So, why is the market (for fabric computing, converged infrastructure, unified computing) still considered so new in the market, yet forecast to be so booming in 2-4 years?

First of all, what we’re talking about here are systems like Cisco UCS, Egenera PAN Manager, HP VirtualConnect, IBM Open Fabric Manager, and a few others. At the heart of each system is technology (sometimes HW, sometimes SW, sometimes mixed) that virtualizes I/O and leverages converged networking.

And why are vendors all chasing this approach? For a number of reasons --

1. It’s incredibly complementary to virtualization: in the same way that the hypervisor changed how SW is abstracted, provisioned, managed and migrated, Converged Infrastructure changes how IO/networking/connectivity is assembled and managed. This gives vendors a valuable set of new offerings, and can tie management of infrastructure to management of VMs – yielding end-to-end abstraction of the entire data center. Roughly as much $ is spent managing infrastructure as it is managing software… to the TAM is huge here.
2. It changes how availability is delivered: By manipulating IO addressing, networking and connectivity, Converged Infrastructure Management can re-provision failed hardware – either in the form of physical servers, or indeed, entire environments. Thus, Converged Infrastructure has the potential to displace a big chunk of traditional clustering software… (nearly a $ billion, if you follow IDC’s estimates)
3. It changes how networks are physically wired and managed: Converged Infrastructure uses fewer IO components (either a single LOM or a single CNA), converged network protocols, fewer cables, and generally fewer switches. This yields a lower CapEx investment, and a commensurate lower OpEx to manage. The opportunity to sell alternative approaches to each of these technologies is immense.
4. Converged Infrastructure is highly complementary to shared storage: the pervasiveness of SAN storage is a major enabler of a more virtual/flexible data center. As physical/virtual servers move, migrate and scale, storage simply follows.  An increasing ratio of servers – especially blades – are being shipped with HBAs, indicating that SAN use is on the upswing.
   

As to evidence that this market is shaping-up, we need only look to the magnitude of investment that Cisco, Egenera, HP, IBM – and even Emulex and Qlogic – are pouring into this market. Methinks we’ll see the hockey-stick shortly.