There is a lot of talk lately about new self-service options for cloud computing and virtualized enterprise IT resources. Nearly every vendor offers a way to deliver on-demand cloud services to users without the need for help from the IT department.

Self-service for the cloud is valuable to both the business user and the IT department, much in the same way that booking a flight reservation online or withdrawing money from an ATM machine are helpful in our own personal lives.

But as many organizations dig deeper, they're finding that not all self-service offerings are the same. There are four basic options currently available:

1. First Come, First Served
This type of self-service cloud capability is similar to the old-fashioned video movie rental process. When you went to the video store, you could only check out a movie if the store had it in stock. Similarly, the first person to request the cloud or IT resources in this case gets the resource reserved for them. But there are no guarantees you'll get the resources you want until you book them. There is also no method for dealing with priorities or conflicts. This kind of self-service is good for IT environments with limited demand and nearly infinite resources

2. The Request Queue
The Request Queue model is very similar to "First Come, First Served" in that it is totally on-demand but also does not guarantee resources. However, it does allow the user to request a resource that is not currently available. That way, when it does become available, it is automatically delivered to the user. This is similar to the Netflix self-service model. You log into the self-service portal and request a particular movie. If it's available, it's shipped out immediately and you're a happy camper. If it's not, it remains in the queue, and as soon as it becomes available, it's shipped out.

However, imagine you are a tester who needs a four-tier SAP environment for your next round of testing. You place a request at 8:00 a.m. Monday morning and are denied because all the resources are already taken. You register your request in the queue via a trouble ticket. Once the environment becomes available, perhaps even two weeks later, it is provisioned and delivered you. Although you did receive the resources in the end, this is an issue: You've been unable to adequately manage your testing time, as you've had no visibility into when you'll actually be able to access the system. And, of course, if users are having problems accessing resources, they are going to pass those problems along to IT.

3. Scheduling System
A Scheduling System seems like it should address the challenges we've described so far. However, if there is no guarantee that the resources will be delivered, this can be even more frustrating for users (and administrators) than even the first two self-service approaches.

An old episode of Seinfeld does a good job illustrating some of the limitations of this approach. In that show, Jerry arrived at a car rental agency with reservation in hand only to find out that there were no cars available. The ensuing discussion centered around the fact that the important part of a car reservation isn't so much the taking of the reservation, but the holding of it...

Simply put, a scheduling system for the cloud is able to take the reservations, but without the ability to manage underlying capacity and priorities, it's not able to hold the reservation, creating no guarantees for the end user. The illusion of the reservation ends up being even more frustrating than the denial of a request.

4. Guaranteed Reservations
The Guaranteed Reservations self-service model is what most IT organizations want. It's similar to what consumers experience with their airline reservations. This type of system allows the end user to reserve the required resources guaranteed, and then access them at that time without the need for help from IT. The IT administrator does not have to deal with conflict resolution. And he or she can also govern how resources are distributed, speeding the delivery of complex resources to users.

If self-service is something you want from your cloud computing services or virtualized datacenter, be sure to ask the right questions so you get the kind of self-service that's right for you and your organization.

• New OpenNebula Self-Service Portal for Cloud Consumers