|
 |
|
General Information
|
Add To My Personal Library |
November 2, 2007
• Vol.29 Issue 44
Page(s) 5 in print issue
|
Keep An Eye On Those Drives: Don’t Neglect Server Disk Maintenance
|
Disks just spin, don’t they? After all, the MTBF (mean time between failures) ratings of modern drives are in the millions of hours. And the IT manager has plenty of other items to look after. But ignore server disk health at your peril. The last thing you want is a critical server failing because of a disk crash. Even if you have RAID in operation, a dead disk will slow performance down to a crawl until IT installs a new disk and the RAID system rebuilds. And in any case, MTBF ratings do not necessarily mean that a given disk will run for millions of hours error-free.
 Defrag Those Drives
Proven over the course of a decade and a half in Windows-based servers, probably the biggest single disk maintenance action that you should do is defragmentation. Over time, server files get pulverized by fragmentation. This results in files splintered into thousands of pieces, and that’s often the reason for a host of woes: crippled performance, slow backups, file access taking ages to load, systems taking forever to boot up, etc. Defragmentation software is available from companies such as Diskeeper (www.diskeeper.com), Raxco Software (www.raxco.com), and O&O Software (www.oo-software.com).
“Renasant Bank performed an evaluation of defragmentation and found that servers with Diskeeper were able to handle, on average, five more workstations than servers without Diskeeper,” says Michael Materie, director of product management at Diskeeper. “During the reliability test, the servers that were defragmenting files automatically had up to 10% higher uptime than servers that didn’t have defragmentation software automatically running.”
Analyst firm IDC estimates the total savings to Renasant Bank is $19,520 per year per 100 servers, with an ROI of 165%.
Image Is Everything
Hamid Azar of Azar PC has been fixing servers and PCs for years. He is a firm believer in the importance of taking drive images as an important aspect of disk maintenance. What this does is takes an image of the entire drive and transfers it to another hard drive. Tools that accomplish this include Norton Ghost by Symantec (www.symantec.com) and R-Drive Image by R-Tools Technology (www.r-tt.com).
“If there is something wrong with a hard drive, you don’t waste time looking for individual files, as it could crash and then you might not find anything,” says Azar. “Take an image of it, and you are safe 99% of the time.”
Go Small
2.5-inch drives have been around for a while in laptops, and they are now emerging as a viable and reliable format for server drives. Seagate (www.seagate.com), Hitachi (www.hita
chi.com), and Fujitsu (www.fujitsu.com) have either announced or released such SFF (small-form factor) drives. The reasons that they have become popular in server environments include their 70% smaller size, resulting in higher performance and 25% less power consumption than their 3.5-inch counterparts. Plus, they deliver industry-leading reliability levels, such as 1.6 million hours MTBF for the Seagate Savvio drive.
“One of the best current server drives on the market is the 2.5-inch SFF drive,” says Henry Fabian, executive director of product marketing at Seagate. “It gives customers a higher-performing, higher-reliability solution that operates cooler for better overall system reliability and operation.”
Fit The Drive To The Purpose
Drives do not equal drives do not equal drives. As noted above, they can be different sizes. Then there are SAS drives, which offer higher performance, and SATA drives, which provide a greater amount of on-disk storage but at lower speeds. And within the SAS and SATA categories, different rpms and features are available.
“All drives are not created equal,” says Fabian. “One of the most important considerations in minimizing your drive and server maintenance is to always ensure you are using the right drive for your application requirements.”
Tiered storage, for example, is becoming more and more common. The highest performance Fibre Channel storage is used for the most critical applications that demand top speeds. However, it’s expensive. So the second tier of storage might be offloaded to SAS drives, while a third tier might exist that consists of SATA drives. Thus, the Fibre Channel environment is reserved for super-fast access, SAS for data of lesser importance, and SATA for the bulk of information that either isn’t accessed frequently or is used by applications that don’t demand the speed of light.
“It is vital to understand what your tiered storage requirements are and to match your performance, reliability, capacity, power, and data integrity requirements to your storage,” says Fabian. “Matching your drive to your application optimizes your processing and business objectives, as well as minimizing your service and maintenance interruptions.”
by Drew Robb
SMART TIP: Use Available Tools
Many operating systems include preinstalled tools that can scan the hard drive, diagnose problems, and fix disk errors and other issues. Investigate what you have available and use these tools liberally. Alternatively, go to the Web site of the hard drive manufacturer and look for any free utilities that may conduct detailed testing of the hard drive. Such tests include a check of the entire hard disk surface, as well as a test to see if the hard drive mechanism was sound or was in actuality about to fail.
Alternatively, server management tools from HP (www.hp.com), Dell (www.dell
.com), CA (www.ca.com), and BMC (www.bmc.com) can assist in monitoring disks and detecting increasing error rates as a forewarning of pending failure. Further troubleshooting gems include Ontrack EasyRecovery by Kroll Ontrack (www.on
trackdatarecovery.com), Undelete by Diskeeper, and hIOmon from Hyper I/O (www
.hyperio.com). If that doesn't work, another option would be to ship the hard drive off to a lab for specialist recovery work. Examples include the Ontrack In-Lab Data Recovery Service and Iomega Data Recovery Services (www.iomegadatarecovery
.com).
Hamid Azar of Azar PC says he has had occasion to use such a service. “I ended up paying $1,700, but it was worth it for me due to what they recovered,” says Azar. “If your hard drive won’t spin at all, a lab represents your last option.” |
EASIEST TO IMPLEMENT: Manage User Quotas
Servers can fill up rapidly, particularly those used as file servers. One time-honored solution is to assign each user a specific space quota. Quota management software automates much of the quota management process. Such programs make it easy to set quotas, send reminders once quotas are being approached, enforce file deletion or archiving when a quota is exceeded, and so on. The use of quota management keeps servers from stalling due to running out of storage. Ideally, you want to never let a disk exceed about 80% of space consumed. Otherwise, performance tends to drop, defragmentation software sometimes struggles, and you run the risk of running out of disk space altogether. |
BONUS TIPS
Disk failures happen early or late. Disk failures don’t happen gradually over time. They actually follow a definite pattern. Most disks either fail early due to manufacturing defects or wear out over time due to heavy usage. So be sure you pay most attention to brand new drives or older ones.
Keep it clean. A simple action that anyone can do is to get rid of unnecessary files from a server. Temporary files, temporary Internet files, recycle bins, and other areas can fill up rapidly with junk. Check them out regularly and discard such files, using whatever utility is available under System Tools. |
|
|
