We were presenting the benefits of Dynamic Provisioning to the architecture group of a very large company recently. The benefits include thin provisioning which eliminates the waste of allocated unused capacity, wide striping which can increase I/O performance by spreading the I/O across a large number of disks, and dynamic provisioning which enables the dynamic allocation of shared virtual capacity as required.

One of the systems architects made the observation that since they were using managed services from a third party, they may not see the economic benefits of thin provisioning. According to the terms of most managed services contracts, the customer pays on the basis of allocated capacity, not on the capacity that is actually used. While the managed service provider could benefit from thin provisioning, it was not clear whether that savings would be passed on to the service consumer. I suppose the question would be the same for an IT shop that was doing charge back to its end users. Do you pass on the savings by charging only for what is used, do you charge a reduced rate for the requested allocation, or do you continue to charge the same rate based on the fact that you still need to manage thier resources in the event that the total, requested, allocation is required?

While they would have to work that out with their service provider, one of the other architects, said that the cost of capacity was really not the most important benefit. To him the benefits of increased performance and dynamic provisioning were much more important. If he could reduce the time to do batch processing or improve his transaction response time with wide striping and be able to provision a new storage request in a matter of seconds from a pool of virtual storage capacity, he would be happy to stay with the same pricing scheme. I suspect that he might even agree to pay more for these other benefits.

Another advantage that can be gained from thin provisioning, even if you continue to be charged for allocated capacity, is the time that is saved in making copies, moves, and distance replication of thin provisioned volumes. Without thin provisioning, the control unit would not know how much capacity is actually used and would need to copy, move, or replicate the whole volume even if only a fraction was used. What is consumed is not only capacity, but, more importantly, time. For instance the time to make a point in time copy of a TB of allocated capacity versus the time to copy 200 MB of used capacity may be a difference of hours versus minutes, (including the use of wide striping).

The value of Dynamic Provisioning is less about saving capacity and more about saving time and providing agility.