Owning Servers is Passé
Cloud computing isn't just for little guys anymore. Both local hosting businesses and MIS-managed computer centers are likely to disappear into the clouds in the next couple of years. Small is NOT beautiful in host computers anymore. Amazon (for example) CAN do it better and cheaper than you.
For years many of us have been hosting our applications with local hosting companies. Sometimes this has meant buying a computer and having it in a cage at the hosting service; sometimes this means renting real or virtual computers from the hosting service. Usually it is necessary to buy as many computers or commit to as much capacity as may be needed to service your worst (or best!) case peak. If you own the computers, you have to replace them when they become functionally obsolete.
Cloud computing as the term is used in this post means running your applications and/or hosting your data in the huge clouds managed by Amazon, Google, Microsoft and others (I've only used Amazon so my examples come from there). When you use the resources of a cloud computing provider, you don't think in terms of specific computers but rather in terms of data storage capacity and virtual servers. Amazon (again, I don't know as much about the others) only charges for what you actually use in terms of storage and Internet access. If you have a peak, you pay for the peak; but you don't have to maintain any idle capacity to allow for peaks. The capacity is just there when you need it. If you have actual applications running at Amazon rather than just data stored there, you do pay per hour for each "instance" of the application including its virtual server which is running but they can be started and stopped dynamically depending on actual load. Small instances cost $.10/hour to run.
Instances consist of any one of a number of different Linux or UNIX flavors or Windows Server 2003 plus the applications you've chosen to use with it. If your applications already run in one of these environments, chances are that you get them running on Amazon EC2 relatively easily. Familiar software like SQL, mySQL, Oracle etc. is already available there in the clouds.
Customer-facing servers are especially good candidates to go into the clouds. Amazon isn't perfect but you're probably doing fine if you have no more outages than they do. Your data is safely removed from your premises and replicated. Your servers keep running even if power is lost locally. You can opt to appear in North American, European, and/or Asian clouds so you can be "near" your users in an Internet sense without having to position servers around the world or paying a content delivery network like Akamai to cache your data near your users. Amazon is connected directly to Internet backbones so data gets in and out more quickly than it's likely to do from your premises or even from most local hosting organizations.
If your organization is spending hard-to-get capital dollars buying servers, that decision needs to be revisited pronto. If you're looking for a place to cut costs, look at the cost of care and feeding for your own servers. How much time are technical experts spending making sure that your servers are running correctly or that strange things haven't happened to the connection between your server and the Internet? How much do you spend on updating the servers? How much extra capacity do you have to buy to allow for peaks in demand? How much extra Internet bandwidth are you paying for because there might be a spike in customer queries? Most of these costs can be avoided with cloud computing.
Business winners and losers: local hosting services that survive will segue into local experts skilled at getting their customers running well in the clouds. You still need the right application for your business; you still need well-designed web pages; if you're not a technical business, you don't want to be creating virtual instances of servers (even though this is easier than managing real servers). In fact, the local experts will have less of a conflict of interest when they're not promoting services which require you to buy more capacity from them.
Dell, Sun, and other manufacturers of server hardware lose. There won't be as many small servers sold. The big guys who big servers have enormous bargaining leverage.
CTOs who give up the glass rooms and move their organizations into the cloud may lose capital budget but they'll gain kudos.
I think content delivery services like Akamai lose because data replication comes bundled with cloud hosting. That's certainly what Amazon's CloudFront is all about.
The cloud providers win. They shares their economies of scale and gain further economies of scale. They get a small piece of a huge volume of computing, storage, and data access. They get to sell auxiliary services like billing your customers for use of your hosted resources. Disclosure: I own a small amount of Amazon stock purely because I believe they are leading the way in cloud hosting (and I like Kindle), but not because of their core book selling business.