Cisco & the Cloud

Many years ago when I was a neophyte techie I worked on a project to fit out a new building in Richmond, Surrey. The plans included our own data center. The building was going to be the European headquarters for a big company and it needed somewhere to put all their servers.

I went on to be involved in many more building projects but this was my first and last co-lo. Before about 1994 nobody was good enough at consistently providing co-lo services, it was a very custom and expensive business, but once that barrier came down everyone moved into rented space.

We’re seeing a similar thing happen now with ‘The Cloud’. The idea of renting time on computers has been around for a while but there have been a number of barriers: bandwidth, security and most of all, price. As these are falling, I believe we’re seeing a similar paradigm shift (forgive the cliche) away from renting your co-lo yet owning your servers to renting your servers. It’s just more cost effective unless you have literally thousands of servers.

I’ve been wondering how long it would take Cisco (and one or two other vendors who are big sellers into the data center such as Sun, Dell & HP) to get into the cloud business. Cloud Computing is a clear and present threat to these guys. The less imaginative ones will continue to try and sell the old way while the future is bright for those who use their manufacturing expertise to build these large cloud co-los using their own hardware & people. Folks like Sun have been providing COD (capacity on demand) systems like this for years with products like their E10K and would do well, they could get back into bespoke hardware that helps the virtualization work more efficiency for instance.

Today ChannelWeb has published rumors that Cisco will introduce a range of blades next year. You can already run Linux on their ACE blades (a nice neat solution for managing large switches) and this is a logical next step. However, mass producing cheap servers is an order of magnitude different from producing cheap switches and I think Cisco may be underestimating the difficulty of doing this well, in particular handling logistics and QA. Both Dell & HP have had to work very hard to stay on top of these issues and keep prices down.

Personally, I think Cisco should buy Sun and Rackspace, hire a few folks away from AWS and get into the Co-lo business at scale before Microsoft, Amazon & Google sew up the market by converting their already massive co-lo investments into rentable spaces.

This will be an exciting space to watch over the next few years.

UPDATE #1: Cisco Enters the Server Business (16 March 2009)
UPDATE #2: IBM in Talks to Buy Sun for $7bn (18 March 2009)