| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Valued Partner News February 2010
Volume 1, Edition 2
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
Greetings!
During the past month or so, Ol' Man Winter has unleashed his awesome power from the East to the West. This provides a perfect opporuntity to talk about the all-important topic of Data Protection & Availability as it relates to disruptive power conditions.
Our goal is to provide you with some beneficial information to help evaluate your important office and datacenter environments that may lack adequate protection for your important business data.
As always, thank you for your business, we value it very much.
~Your Partners at Partners Data Systems
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
RAID 6, Protect your data from bit errors...
It's common knowledge today that RAID-6 protects you from a double disk failure, unlike RAID-5, but what you may not be aware of is that RAID-6 also protects you from data corruption due to bit errors.
(The following is excerpted from an HDS whitepaper)
Bigger Drives Bring Bigger Risks
 While new 1TB (and 2TB) SATA drives have entered the market, many system vendors are only now in the process of adding them to their product portfolios. With these larger disks comes an increased risk for data loss resulting from a failed drive rebuild in a RAID system.
Let's pause for a little background. If a drive fails in a RAID group, the other drives in that group must rebuild the failed drive from scratch. For a rebuild to succeed, all of the data on all of the sectors of the healthy drives must be readable. If any sector is unreadable, the rebuild work hits a wall and comes to a stop (or as with many RAID systems continues the rebuild with data corruption).
At that point, you have lost data or you have corrupted data if you are using RAID-5 or some other single-parity system. A RAID-6 configuration gives you a second chance if the drive group encounters an unreadable sector during a rebuild. This second chance comes with the second parity drive that is there to correct the errors in the
process of rebuilding the RAID group.
So how likely is it that a RAID group would encounter unreadable data? It is more likely than you may think. Here's why: When a drive's read-and-write head reads data on a disk, it will rarely make a mistake. Any mistake is known as a "bit error." On average, bit errors occur at a predictable rate. Drive manufacturers can tell you the bit error rates for their products. This rate refers to the average number of bits of data that are transmitted incorrectly when data is being read on a disk. Today's server-grade SATA hard disk drives have a standard bit error rate of 1 unrecoverable read error per 125TB of data. This means that for every 1TB hard disk drive read, there is a 0.8 percent chance that one sector will be unreadable.
While bit error rates are extremely small, they become more significant with larger drives and larger RAID groups. That is because the more bits that are read on a disk, the more chances you have for bit errors - and the subsequent failure of an attempt to rebuild a drive.
· With a typical RAID-5 configuration using 1TB SATA drives in an 11 drive RAID group, the probability of data loss due to the combination of a disk failure and the bit error rate is 7.7 percent!
· With a RAID-6 system of the same size, the probably of data loss from the combination of a disk failure and the bit error rate is virtually 0 percent (even with larger 2TB drives).
In other words, when you lose a disk drive in a RAID-5 configuration, there is a significant chance that you could also lose data.
Partners highly recommends RAID-6 for most applications (and RAID 10 for high performance applications. RAID 10 provides nearly the same protection as RAID-6 but provides significantly less usable disk space).
Even though the research is from HDS, this information applies to all RAID systems. The larger the disk drives the higher the risk. RAID-6 Whitepaper
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
