I have a friend who is an software architect in a large Wall Street bank, or lets say, remnants of a large Wall St bank! One of things he mentioned to me a few years back was that whenever their team wanted to deploy a new application the biggest headache was finding the power to deploy additional hardware in the already power saturated data-center.
These days there is a lot of talk about private clouds. Some companies (a lot of them are start ups) are building cloud solutions for the datacenter. There are a lot of hurdles in the path of these companies. Lets look at a few:
1. What happens to the volumes of code that already exist in that datacenter? Is there a migration path for these applications to use the private cloud's RESTful or other APIs? One can easily imagine that all new applications can be written using the cloud APIs. However, given the current economic situation and the oft used mantra "if it ain't broken, don't fix it" it would be really hard for existing applications to be ported to a cloud model. The best that could happen is that these applications run on virtualized resources which may already be happening.
2. What happens to the existing service delivery model? The customer service, the ticketing system to the fix hardward, software problems, etc. are all based on the traditional model of the datacenter. All of this wold need to be overhauled when the datacenter becomes a private cloud.
3. The management software to manage existing datacenters is something companies have invested quite heavily in. They have also trained their man power to use these enterprise management software. Either the private cloud is able to work with these management applications or else the switching costs will be very high.
All of this will dissuade existing data centers from switching over so easily. The private cloud provider will need an army of service professionals who can come on board and make the transition smooth for the enterprise customer. That makes me think small companies are not in a position to make an impact with a private cloud solution. Large companies like Sun Microsystems, IBM, Dell, HP are better positioned to affect such a change.
The other way for private clouds to make a foray is to be able to disrupt the dynamics of a traditional data center. One way might be to have the private cloud support an aggregation of all the services that people are looking for these days: the collaboration tools, the social networking apps, the build/deploy services, etc. The key is to provide these services in a way that lowers the learning curve. The other key thing private cloud providers need to do is keep on harping about the improved data center utilization of the new architecture. Also the private cloud APIs need not be private as such. These could be seen as fronting public cloud providers too. That way when the data center does run out of hardware resources some applications could use public clouds without the applications being affected in any way. Applications that don't need to run in-house from a security/privacy point of view can be outsourced to the public cloud.
Bottom line, private clouds are viable but only for new applications. Existing applications can move to a private cloud one application at a time. Its like the early days of the web when IBM deftly coined the term "e-business" when all it did was move Mainframe applications to the web one at a time. In the process they sold lots of mid sized Unix Servers and huge service contracts.
