Sun CEO misses the boat on datacenters

Jonathan Schwartz, Sun’s CEO, asserts that datacenters are doomed in his latest blog post, asking “Why bother with datacenters at all?”

As the owner of a web hosting company with a significant amount of datacenter space, owner of several Sun servers, and as a former Sun employee, my take on this is “I’m really glad I sold my Sun stock.” Schwartz clearly has no idea what his own customers are doing with the servers Sun sells them, or why Sun is really struggling in its core “big iron” market. He must have not taken a datacenter tour lately or priced out space in a datacenter (hint: prices are increasing rapidly as space fills up.)

First, I agree with Jonathan on one key point (apparently the only point in his blog entry that is based on actual analysis.) He says, “At least from our internal analysis, the availability of IT infrastructure is inversely correlated to foot traffic. The more people allowed in a datacenter, the more likely they are to kick a cord out of the wall, break something trying to fix it, or just bump into things…. As the best systems administrators will tell you, the most reliable services are built from infrastructure allowed to fail in place, with resilient systems architecture taking the place of hordes of eager datacenter operators.”

Right. Schwartz is correct in this point — people in datacenters are generally not a good idea because human error comes into play much more often in that scenario. But to extrapolate that this means datacenters are not needed is absurd. This in the face of news that Google is building a huge datacenter in Oregon and Microsoft is building a huge datacenter in Washington. And, if you ask around here in the Valley, you’ll find new datacenters opening all the time.

So why will datacenters continue to be relevant? Simple — economies of scale. Building out 10Gbit of pipeline to the middle of nowhere is expensive. It’s also more expensive to build out 1 T1 (1.5Mbit) to 100 locations with 1 server in each location than it is to build out 4 redundant 100Mbit connections to 4 major datacenters and putting 25 servers in each location. Not only is it cheaper, but you get more bandwidth and better connectivity to every server in the datacenter scenario… and much more flexibility should you find yourself at max capacity in a year and needing to build out more servers, space, and add additional bandwidth.

Power, space, and bandwidth (connectivity) are all cheaper when purchased in bulk quantities… and they’re all cheaper when you’re next to the power plant and next to a SONET (fiber) ring. I agree that fewer datacenters will be built out in downtown areas in the future. Datacenters will be built where there are 3 things that come together:

1. Cheap-ish labor. Remote hands/reboot monkeys can be there 24×7 to provide security and do anything that needs doing in the datacenter for $10/hour or less.
2. Cheap power. Google building their datacenter in Oregon next to a hydroelectric dam, for instance… and a couple huge datacenter installs in Las Vegas, where power is cheap. Here in the Bay Area, there’s a whole street full of datacenters coming online in Santa Clara, where there is a co-op providing much cheaper power than PG&E can provide.
3. Availability of large amounts of bandwidth. What amazes me the most about Schwartz’s latest post is that Sun’s entire motto used to be “The network is the computer.” Look at Google releasing an office suite, for instance. What about salesforce.com? Web-based email? Where does Sun think these services come from? As the number of web-based services continues to grow, and their quality continues to increase, more and more companies will be subscribing to these services. These services are particularly tailored toward small business, a market that hires 70% of the employees in the U.S. but a market that Sun has never understood. Sun was always about Big Business. But small businesses drive the market forward, and most innovations come from the small business side. If Google et. al. can make office suites and email over the web work like email and office suites do today on our PCs, a huge market will open up.

What is behind that market? DATACENTERS!

What boggles my mind is how I know this and how Schwartz, CEO of Sun Microsystems, didn’t seem to get that message. I know this because we have hundreds of customers who are using our datacenter space to provide services over the web that would normally be relegated to a back-office server somewhere. Companies don’t want those back-office servers any more. They don’t want to pay IT staff to maintain them and they don’t want everything to be offline if one crappy desktop gets its power cord kicked out (exactly the same scenario that Schwartz mentioned in his blog post.) That’s why they’re putting servers in datacenters! We get calls like that so often that I’d say there is easily a $10M/year (yes, ten million dollars a year!) business in providing that sort of outsourced IT to companies. Now here’s the kicker. I said $10M/year. I think there’s a $10M/year business in providing that to companies in Santa Clara county alone. Providing it nationally? It’s worth billions! No question about it. Where do you think IBM Global Services gets so much money?

I have no idea what makes Sun so blind to this market. Perhaps it’s because they tend to ignore small businesses (as mentioned above.) It’s the smallest businesses (fewer than 10 people) that are demanding this. They are screaming for someone to come in and put their Exchange/Novell/groupware/file sharing services/website on a few redundant servers in a datacenter (and then back it all up to another datacenter) and then give them a private leased T1, T3, etc. back to their offices, and a smallish connection to the Internet to serve files out, serve a website out, and serve outgoing email. And they’re willing to pay thousands of dollars a month to have someone do it, because paying a company $5000-$8000 a month to do this is cheaper than hiring one or two IT people! The companies that successfully pull of this migration will make millions and millions of dollars.

So why not Simpli? Well, we’re in a slightly different market. But we’re still looking into it, because the demand is so high. It will take a lot of engineering talent to pull it off. And it takes a lot of staff resources to set up this infrastructure. The big question for this new company owner is: why use Sun servers to power your huge datacenter when you can use cheaper Intel or AMD-based servers running whatever you want? The Web 2.0 industry already knows this, and it’s why Sun missed the bandwagon there too. I’m amazed that Sun doesn’t see this, but with those “big business” blinders that Sun has, they would never acknowledge this market. Someone will, though, and Sun will become even more irrelevant than it is today.