The best way to simplify the complex undertaking of data storage is to switch to a storage utility model, according to Ron Lovell, storage practice director for Greenwich Technology Partners. The transition can be complex, but the end result will be a more responsive data storage management system coupled with the appropriate equipment in your infrastructure. In some cases, the gear would simply be cheaper—some storage tasks don’t require all the bells and whistles—and in others, the same equipment would be running at a higher capacity.
Lovell, speaking at CeBIT America recently in New York City, said that reducing and consolidating storage and integrating it into a utility model leads to better return on investment and, if done correctly, better performance. This is important in an era in which the demand to store data—because of the explosion of new information, new data warehousing requirements in some industries, and the attention paid to preserving records in the post-9/11 world—is growing among companies of all sizes. The following is more on what Lovell had to say about developing an enterprise storage utility model.
The state of affairs
There are a host of concerns about the state of enterprise storage. Clients, Lovell said, are having difficulty keeping pace with recovery time objective (RTO) requirements. Also, the economy is putting pressure on storage, which can be seen by nontechnical executives as a good place to cut costs. Competing requests by business units and applications make deployment without comprehensive management software and procedures difficult and likely to lead to an inadequate infrastructure. Finally, direct server-attached, storage-area network (SAN), or network-attached storage (NAS) often are implemented in an uncoordinated manner. In many cases, the gear is only working at 25 to 35 percent of the available capacity.
The general confusion represented in the above scenario has predictable results: IT organizations tend to buy high-end devices that are unnecessary. Also, most of the data that is stored could just as adequately be housed on less expensive machines. Greenwich estimates, according to Lovell, that 90 percent of stored data is untouched four to six weeks after it is stored. In short, having gear in place that can instantaneously retrieve data is unnecessary in a vast majority of cases. Another chronic issue is that companies often pay for features they don’t know they have.
Types of consolidation
These are complex issues for enterprises and even storage service providers (SSPs) to deal with, and Lovell laid out five types of storage consolidation to help:
- Facility consolidation, as the name implies, relates to reducing the number of data centers.
- Basic rationalization focuses on reducing the number of servers running the same application. So, for instance, a job can be done on a two-server cluster instead of five discrete servers.
- Complex rationalization is the same concept as above, across more than one application. For example, three applications could be condensed from 15 servers to three server clusters.
- Function consolidation relates to finding ways of eliminating applications that duplicate functions. An example of function consolidation would be the merging of two online banking Web sites into one.
- Server virtualization is the process of making multiple physical servers act as a single unit.
The first step toward rationalizing the complexity, according to Lovell, is to work with these consolidations in a manner that offers storage as a utility to customers. This can alleviate complexity by creating a coherent context into which the various consolidation tactics and strategies fit. The benefit is discipline. A utility solution employs far more stringent planning, oversight, and management than ad hoc solutions. A utility solution will be more closely tied to actual need, both in the type of storage and the capacity that is planned. Another benefit is that a higher percentage of the storage facilities can be managed by a single full-time employee. The end result is that the total cost of ownership (TCO) shrinks and performance improves.
Storage as a utility
The idea is to create multiple offerings that include primary and backup functionality. A business model should be realistically priced and include planning, deployment, provisioning, and management functions. This should be accompanied by a storage management framework that also controls planning, deployment, provisioning, and management of the storage, which is now configured as a utility. A cross-functional storage management team should be created and the appropriate application development folks should be made aware of, and encourage, use of the new functionality that the revised infrastructure offers.