About two years ago I made the switch to disk-to-disk backups within my own organization. The volume of data I was backing up each night had gotten to the point where it exceeded the capacity of a tape. Even multi-tape backups had become impractical with my system.
At the time, the decision to switch to disk-to-disk backups was purely financial. For me, it was significantly cheaper to implement a disk-to disk-backup solution than invest in a tape drive that could accommodate all of the data I was backing up each night. Still, the disk-to-disk solution was far from inexpensive.
Back then, terabyte hard drives were still about $400 each, and when you combined the cost of several hard drives, a RAID controller card, an external chassis, a power supply and all of the various cables that I needed, you can see why the costs were significant.
Recently, I needed another RAID array that was about the same size as the one I just told you about. I knew that the cost of hard drives had come down a lot in the last couple of years, and I was hoping to find some other ways of cutting costs without sacrificing performance. As I began to research the problem, I learned something really interesting.
As you can imagine, one of my biggest expenses associated with building the RAID array was the controller card. Fortunately, if you want to build a RAID array these days, you may not even need a controller card. Intel Corp. has actually begun integrating RAID controllers onto many of their system boards. The crazy thing is that few (if any) of the computer manufacturers that use Intel system boards even advertise this feature.
Intel calls its integrated RAID controller Intel Matrix Storage. The idea is that system boards featuring Intel Matrix Storage are equipped with six SATA ports. You can use these ports for standalone drives or you can create a RAID array consisting of two or more drives. Some laptops also feature Intel Matrix Storage, but typically only offer two SATA ports, which allows you to implement RAID 0 (striping) or RAID 1 (mirroring).
The system board that I bought for my particular project also offered two EIDE ports, so I was able to install an EIDE hard drive and use it as the system's boot drive. This allowed me to free up all of the SATA ports for a RAID array that I used solely for data storage. As of right now, I'm only using four of the SATA ports, but it's nice to know that I have two more ports available to me if I need them.
One thing to keep in mind about Intel Matrix Storage is that, by default, all of the hard drives are treated as standalone drives. If you want to create a RAID array, then you're going to need a software application that Intel calls Intel Matrix Storage Manager. You can download the software for free from Intel.
Having an integrated RAID controller can greatly reduce the costs associated with disk-to-disk backup technology. Of course rapidly declining hard drive prices don't hurt either. One thing to keep in mind: Although disk-to-disk backups offer much more flexibility than tape-based backups, they lack portability. If a disk-to-disk backup is your only backup mechanism, and you are performing the backups onsite, then a fire, flood, hurricane or other disaster could destroy both your primary copy of the data as well as your backup. As such, it is still important to periodically perform a traditional backup that you can ship offsite.
Again, different companies have different needs, so what works for one company may not work for another. In my case, I perform a full backup once a month using a read-only DVD. I then store these DVDs offsite. My disk-to-disk backup solution takes care of backing up my data throughout the day, but should the unthinkable happen, I should never lose more than a few weeks worth of data.
This is the solution that works for me, but for many companies making a permanent backup once a month would be completely unacceptable. I know of several companies that use a disk-to disk backup solution to back up data on an ongoing basis throughout the day and then perform a traditional backup at night. This type of solution tends to be expensive to implement, but it allows you to recover the maximum possible amount of data after a disaster.
ABOUT THE AUTHOR
Brien M. Posey, MCSE, has received Microsoft's Most Valuable Professional Award four times for his work with Windows Server, IIS and Exchange Server. He has served as CIO for a nationwide chain of hospitals and healthcare facilities, and was once a network administrator for Fort Knox. You can visit his personal Web site at www.brienposey.com.
This was first published in May 2008