Users stand to benefit greatly from storage resource management (SRM) software, which allows administrators to see which applications are using what storage. By being able to map applications with back-end systems, IT managers can create better cost-to-value matrixes and develop accurate, functional chargeback models.
This pent-up demand for better infrastructure insight was illustrated at the recent Storage Networking World conference, when a poll of audience members revealed that 27 percent of them considered SRM their most important concern, while 21 percent cited integration with other management tools as their third-most-pressing issue.
The problem is, it's not happening.
SRM vendors have yet to produce comprehensive products that offer these key capabilities, and as a result, users have been left to assemble their own homegrown best-practice solutions. Unfortunately, these one-off solutions are largely incompatible with commercially available SRM offerings when it comes to developing chargeback models, which remain largely based on allocation.
Making matters worse for users, many companies view this kind of resource mapping as a "nice to have" as opposed to a must-have capability because of the complexity involved with implementing it.
And while the emerging SMI-S management standard is lauded by users and vendors alike for its potential, in the eyes of some experts, when it comes down to building the sophisticated systems required to enable heterogeneous mapping applications, SMI-S is still literally relegated to the bottom of the stack.
Home-cooked best practices
Paul Strong, a distinguished research engineer at eBay's research lab, characterizes the situation by saying, "To move forward, we need to be able to map resource consumption with business value delivered." He goes on to note that accomplishing that goal entails tying the value of business processes, workflow and user interaction to sets of services and applications. Regarding the ability of commercially available SRM software to perform these functions, he adds, "Typically, you can't do this."
EBay is filling this void by creating its own SRM best practice through integrating some of its existing vendor tools. This has enabled the online auction service to map some applications down to specific databases residing in specific clustered server environments on large arrays. As Strong notes, however, this still only creates limited cost-to-value comparisons.
Less is definitely more for eBay, which relies on a "cookie-cutter" approach to IT infrastructure based on deploying the absolute minimum number of components, such as arrays, and processes, such as provisioning LUNs, volumes and operating systems.
"In order to make our environment scale, we standardize on a small set of standard processes that become well understood so we could have better leverage and strength in those relationships, so if there are problems, we can resolve them more easily," Strong notes. "Obviously, the fewer components we need to worry about, the fewer sophisticated management tools we need to care about."