Article Source :: Processor Magazine (www.processor.com)
| Six Quick Tips |
| General Information | |
| July 18, 2008 • Vol.30 Issue 29 Page(s) 26 in print issue |
|
Better Storage Management Make The Most Of What You Have |
|
Better storage management is about better use of existing storage. With the right techniques, SMEs can reclaim or reserve existing storage, replicate and restore data fast, and run applications without disruption.  Advanced Architectures While RAID architectures are good for small disk storage, according to Geoff Stedman, senior vice president of products and marketing for Omneon (www.omneon.com), large disks are another matter. If an SME loses one disk in a RAID array, it can still access the rest of the data while the disk is rebuilding. But because drive storage capacity has increased at a faster rate than have processor speed and performance, it takes many hours for today’s processors to facilitate a very large drive rebuild in a busy storage system, according to Stedman. He says, “This creates a vulnerability because a second drive failure during that rebuild process could mean a loss of data.” If this vulnerability is a concern, you should consider newer storage systems that replace RAID and offer faster drive recovery times, as well as increased bandwidth to the remaining disks and data stores, according to Stedman. One such system, for example, offers data redundancy by splitting up the data into slices among nodes on a grid. With this type of system, losing one large disk for a period of time doesn’t threaten the data, and the grid architecture can replicate it from the failed disk more quickly than RAID can rebuild a drive. The network shares the bandwidth requirements for data recovery, avoiding bottlenecks, which speeds the replication process, Stedman notes. Virtualize For Nondisruptive Data Migration A virtual hardware platform enables data migration into storage without disrupting the application that is using the data, according to Hu Yoshida, chief technology officer at Hitachi Data Systems (www.hds.com). In a virtual hardware platform, the application reads and writes data to and from a cached image of the storage volume, Yoshida explains. There are caches of the old virtual storage volume and of the new, migrated virtual storage volume, and the platform writes data from the application to both images so that both the migrated data and the old volume are accurate and available for the application. Whereas software-based data migration interrupts data reads and writes, slowing the application and costing the enterprise time and money, this hardware process copies data without any effect on the application, according to Yoshida. Save Storage Space With Thin Provisioning A virtual hardware platform can also help save storage space. To illustrate, an enterprise may ask its storage provider for 10TB of storage space when the immediate need is only for 1TB, explains Yoshida. The enterprise then saves its original data on that storage and makes copies of that original data to work from, which takes up still more storage. The enterprise often copies all the unused space from that 10TB of storage onto these data snapshots, as well, says Yoshida, which wastes additional space. In order to avoid unused space and copies of unused space, then, enterprises should use thin provisioning. With thin provisioning, that enterprise would request 10TB of data storage from the provider. But, when they start to use it, they only get the actual amount of storage space that they write data to. That way, when they make a copy of the data to work on, they only copy the data that actually exists, not the whole 10TB, so the copy is smaller, using less storage. Reclaim Storage Enterprises should reclaim unused storage and return it to the free pool of available storage, according to Ashish Nadkarni, consultant for GlassHouse Technologies (www.glasshouse.com). Best practices for reclaiming storage include investing in good resource management software to manage hosts, according to Nadkarni. Many people choose not to run the host agents provided with this software, but these host agents provide valuable information for the storage administrator. With them, the administrator can see underutilized drive partitions and reclaim them. Another best practice for reclaiming storage is good end-to-end communication between your application teams and your storage owners so that the storage team gets a better sense of storage forecasts and systems rotations from the other teams, according to Nadkarni. You should know the systems rotations—when a server or set of servers is being retired or decommissioned—so you can reclaim the associated storage, Nadkarni emphasizes.  by David Geer |
About InteleNet
