Delving into dynamic server provisioning

Dynamic server provisioning is an area of management that’s certainly one to watch if you want to efficiently use the servers, processing power and storage throughout your IT infrastructure.

Picture this: Your application server is hit by unusually heavy traffic due to a special promotion your marketing department is running. Although the traffic load is temporary, you still must provide excellent service to your customers.

Some IT organizations have pools of bare spares (servers that are not in use but are available) that can be quickly provisioned with the correct operating systems, configurations and server and application software, and then pulled online to help handle the barrage of traffic. While this can be accomplished manually and with some automation today, there’s much potential to server-provisioning technology being applied more broadly.

Virtualization is one tool. The basic premise of virtualization is to create a pool of resources that can be used and returned as the demands on IT infrastructure ebb and flow. Rather than sending a specific job to a specific machine, the job is sent to a virtualized pool, and it’s run wherever the performance will be optimal. While this concept seems easy, it’s more complex than you might think.

It’s not simply a matter of randomly sending out “jobs” to run on servers that have excess capacity. The loading and performance depends on the other applications that are also running on prospective servers. For example, if an application is sent to the same server that serves up data for a data-intensive application, performance could be hurt. So consideration must be given to what else is running on a system, as well as the timing dynamics with the application. Does it run in the middle of the night? Or does it run at the same time that the other application that requires a lot of resources?

Next, not all resources are alike. The costs of a resource, the technical capabilities of the resource, and the performance of the resource must be considered. For example, in the storage world, the type of storage media used has an associated cost, as well as specific characteristics that may affect the decision to select one over another for a specific use. While tape backup is cheaper, it can be slower to retrieve the information compared to a disk backup.

Another consideration is integrating the provisioning system with other management systems. Dynamic provisioning can’t be done effectively or efficiently in a vacuum. Tivoli’s recent acquisition of Think Dynamics adds an “orchestration” layer to tie together different provisioning systems.

There are also the automated aspects of provisioning that affect the operational piece of the equation. Products such as Opsware, Veritas’ OpForce, HP’s UDC and others are addressing this issue. Data centers and organizations with lots of servers will benefit from automated provisioning.

Where are you, users, with regard to dynamic server provisioning? Although it’s in the press and there has been a lot of activity from the vendors, I’d like to hear what you think and are doing in this area. Are you using tools like this? Are you planning to use these kinds of tools or not? And how soon? What is compelling about it, or not compelling? Please send your responses to me at mailto:Rasmussen@emausa.com >.