Three Almost-Revolutions?

Today, I want to offer three examples of stuff that could be revolutionary but on the other hand could become a yawn.  They illustrate how complex this market has become, and how delicate the balance is between a romping success and an embarrassing failure.  The order here isn’t significant!

Broadcom launched a 200G switching chip that could be an element in the opening salvo for startup or at least new-entrant competitors in the data center and cloud switching market.  The fact that the company had previously announced a 100G network processor means that Broadcom can now supply the chips needed for a cloud switch based on the OpenFlow standard.  Interestingly, that particular slant isn’t being talked about much.

In an OpenFlow switch, all you really need to have is a forwarding table and some lightweight logic to frame control-plane exchanges when a packet is presented that doesn’t have an entry.  Thus, you could build a pretty big fabric with a bunch of fairly dumb chips.  Since OpenFlow doesn’t have any discovery (no spanning tree for example) you can drop the distinction between ports and trunks in a functional sense.  The use would determine the classification, in short.  You could have any number of transit options exiting a switch, any level of meshing.  This could create flatter topologies without requiring a single fabric or virtual switch.

This is the real risk or value (depending on whether you’re a vendor or user, respectively) of OpenFlow.  In my view, too much time is being spent talking about OpenFlow and SDN as a supplement to current switching/routing.  There’s not enough room there to add value; it’s another tunnel protocol in essence.  The question is whether somebody (or somebodies) will deploy OpenFlow switches in applications where connection control is more important than adaptive discovery.  That will be true inside CDNs, inside clouds, inside data centers.

Another story of maybe-revolution is coming from RIM, who is rumored to be prepping a position more reliant on HTML5.  Such a move could have a couple of advantages for RIM, foremost being that it could be played against the big incumbents in the mobile space, the very people who have unseated RIM from its former smartphone high chair.

Which is where the cloud comes in.  I still believe that the real winning position in the mobile broadband world is the behavioral apps hosted in the cloud.  RIM has mail server capability that could in theory be broadened to hosting, and hosting plus HTML5 could create super-apps that would run in smartphone, tablet, and even PC browsers.  Might that be enough to save RIM?  It might have been six months or more ago; it may be too late now.  Nothing short of naked aggression in deployment and positioning of this sort of story will get them back on track at this late date.

Finally, IBM’s Mobile Foundation portfolio is a fairly sophisticated approach to integrating appliances into business operations and building next-gen applications that couple worker and IT in a totally different way.  I’ve pointed out before that the value of the cloud doesn’t lie in doing what we already do in a cheaper way, but in doing what we can’t now do but would like to do.  Obviously there’s a hosting/applications dimension to this, but also an appliance dimension.  That’s what I think Mobile Foundation is.

The heart of Mobile Foundation is a developer environment that lets people write apps that can be ported without change across the range of popular enterprise mobile platforms (Apple, Android, RIM), but the Foundation also includes tools to integrate the mobile apps with current IT and with the cloud.

The thing that makes what seems to be a clearly strategic idea into something that could be much less than that is the fact that IBM links this stuff to the tired old media buzz-concepts rather than the notion of mobile/behavioral empowerment (where it belongs).  They talk about BYOD (old), they talk about how two-thirds or some number of enterprise CIOs want to do more with mobile (older) but they blow by the whole notion that business process re-engineering starts where the business meets the IT process, which is the worker who’s out in the world.

Ten years ago, the real stories behind these maybe-stories would have been shouted by vendors, media, and analysts from the rooftops.  Why not now?  Until we answer that, we won’t understand how to create an IT revolution or a network revolution, so we’ll only have one by blundering into it.


Leave a Reply