Choksi decided to go with a complete rip and replace of their infrastructure. But this would require the shutdown of all the company's applications, so Choksi convinced the operations to give him a 24-hour window of downtime.

The Symmetrix and the Clariion arrays at the primary datacenter were consolidated to one DMX array. The Symmetrix array at the DR site was migrated to another DMX array. The fiber-channel (FC) switches were also upgraded to fully leverage the 4Gbps throughput now available with the newer DMX platforms. The new arrays were populated with the required disk capacity to last for the next 24 months.

In addition, there are certain codes embedded on each disk that need to be upgraded when new software is introduced or maintenance takes place. In order to update such codes, the entire array had to be brought down. Now, selected spindles can be shut off to carry out an upgrade without holding up the entire array. This is crucial as each array runs about 200 spindles, and ensuring the high availability of mission critical applications while allowing seamless scalability is paramount, Hill says. "We have scalability built in for the next 24 months. At that point, we will have to add more disk space as our apps may grow. We are also trying to get into things like managing application performance at the user level. In addition, we have not yet done much work in terms of understanding the quantum of electrical usage while shutting off spindles. This would be of amenable use when we would want to embark on a green IT initiative," says Choksi.

That said, he knows that there's plenty in the here and now that could be taken advantage of. "People tend to utilize what's best for them and leave out a lot of other things they don't have time to utilize. We need to leverage our SAN ecosystem deeply and really push the throughput boundaries that it offers," he says.