Database and storage management: new storage-management products reduce administration and accelerate transactions - Storage Networking

Computer Technology Review, Jan, 2003 by Calvin Hsu

Currently, RAID storage is a widely used means of protecting data at the storage level. With the data redundantly distributed across multiple disks, the enterprise is protected against the malfunction of one of the drives. If the array fails, however, there's potential for big trouble.

Rather than relying on a single RAID device for backend database protection, a better choice is to distribute the redundant copies of data across completely separate disk arrays, on separate power sources, perhaps even separate floors or buildings. Storage hardware with this capability tends to be cost-prohibitive to many who need it.

That was the case with Mark Moroses, senior director of technical services at Maimonides Medical Center in N.Y., one of the country's largest teaching hospitals. "We looked at the typical assortment of high-end arrays from major vendors. But we couldn't find the right combination of features and price that met our needs and budget."

There are also server-installed software and hardware products that can mirror every database update across two different storage devices. However, maintaining components on each server adds to the workload and uses server resources to create the redundant copies.

As a result, organizations like Maimonides are selecting network-level management products that offload mirroring from servers, while maintaining the ability to replicate valuable data across disparate, lower-cost hardware. In fact, Maimonides was recently given a prestigious industry award for "industry-best storage reliability." According to Moroses, "the feature set was right, including support for a wide range of disk devices. This enabled us to buy some quality mid-range disk without sacrificing remote mirroring, high availability options, etc."

Automatically Accommodate Growth

Another headache for database managers and users is the downtime--and related stress--caused by disk space shortages. Administrators must search each server for free disks, incur downtime to remove and then re-install those disks, re-cable and re-configure volumes. This situation is so painful that database owners request an excess of storage to postpone the event, and in turn, the storage buyer incurs added expense. This chain reaction of over-provisioning makes it easy to see why industry analysts estimate that up to 70% of storage capacity is wasted.

Conseco Finance, one of the largest consumer finance companies in the nation, is very familiar with this type of capacity reallocation problem. Under good conditions, moving storage took four hours. In more complex scenarios, up to four days of planning and configuring might be required. An estimated 40% of application owners' time, plus labor from dedicated storage administrators, was spent dealing with these issues.

To help firms like Conseco Finance, storage suppliers and software developers are either delivering or planning automated provisioning functionality. Ideally, the feature provides database applications with just the right amount of storage at exactly the time that it is needed.