The storage industry has been busy trying to remove the bottlenecks caused when an application or a hypervisor needs to access data on a storage system. Scale-out and flash-dependent architectures now dominate the data center landscape. While these technologies have started to improve bottlenecks, there is one problem that we haven't even started to solve: the resources, namely time and people, it takes to adequately manage a storage system.
Storage systems are getting more intelligent and can take automated actions based on conditions or policies, but at some point even the most intelligent of storage systems needs a human to get involved to override the automation or simply create the policy in the first place.
It's not that your average storage administrator isn't working hard enough or doesn't have the skills. The problem is that there are so few of them to go around and, with tight IT budgets and a shortfall of available IT personnel, this situation is going to get worse.
[ For another perspective on storage challenges, see Root Causes Of Today's Storage Problem. ]
The primary culprit is that storage is growing rapidly on two vectors: capacity and performance. The capacity vector should come as no surprise, but the fact remains the more storage you have, the more it takes a human to make decisions. To a large extent, as we discuss in our article "Orchestrating Perfect Storage Provisioning," automation systems can lower the administration time required by storage provisioning requests.
The second vector, performance, requires even more of a storage administrator's time. Tuning storage performance on the storage side of the equation is relatively straightforward, but making sure that applications meet their performance service level agreements (SLA), written or implied, is the bigger concern. The application owners each have their own needs and demands. In an environment where SLA is implied, or worse is stated "as fast as possible," meeting performance expectations of the application owner is almost impossible.
To solve this problem, storage administrators are going to have to safely delegate some of the workload. As we discuss in our recent article "What is Multi-Tenant Storage?" the storage administration process has to be granulized in such a way that it can be distributed to the application owners themselves.
Multi-tenancy allows each business unit to divide up its storage resource allocation as it sees fit, or to "buy" more from the storage team. The application owner is probably the most qualified to make these decisions anyway and likely can do it quicker than the storage administrator. Finally, multi-tenancy allows the storage administrator to have oversight but not need to get involved in day-to-day business line decisions.
Storage hardware and software will continue to get more intelligent and automated, but at the same time we need to be able to scale the storage administrator. Given the current constraints of data center reality, a logical way to solve that problem is to make storage management more granular. A multi-tenant storage system allows this granulation to happen without risking data or performance.
Follow Storage Switzerland on Twitter
George Crump is lead analyst of Storage Switzerland, an IT analyst firm focused on the storage and virtualization segments. Storage Switzerland's disclosure statement.
From thin provisioning to replication to federation, virtualization options let you reclaim idle disks, speed recovery, and avoid lock-in. Get the new, all-digital Storage Virtualization Guide issue of Network Computing. (Free registration required.)
