Chapter Review

     

The disk drive is still the most unreliable component in our computing systems. RAID Levels introduce technologies that allow us to protect our data against the failure of a single disk as well as technologies that will hopefully improve the performance of the whole IO subsystem by utilizing more than a single disk mechanism to retrieve our data. The RAID levels that are commonly implemented are RAID 0, 1, 0/1, and 5. They offer a level of compromise between performance and high availability. The thing to remember about implementing a RAID solution is that the premise behind using RAID is normally for high availability reasons; most people don't consider RAID 0 as a true RAID level, as the loss of a single disk results in no access to any data in the logical drive. The choice of RAID level will either be at a cost to capacity (RAID 1) or to performance (RAID 5). Some solutions will give you overall better performance than single disk solutions but at some cost to usable capacity (RAID 0/1). The choice is yours.

A single disk solution is commonly known as a JBOD (Just a Bunch Of Disks) whereby we may implement a software RAID solution on top of the underlying simple disk technology. More and more these days sophisticated RAID arrays are utilized which offer high availability (or even fault tolerance) and high performance through utilizing a myriad of hardware, high-speed processors, and a vast amount of cache memory. In such a situation, implementing a software RAID solution on top of the hardware RAID solution makes no sense; you are wasting valuable server CPU cycles when the hardware solution is doing it for you.

IMPORTANT

Where possible, ALWAYS implement a RAID solution in hardware, i.e., a RAID array. Software RAID solutions are expensive in terms of server CPU cycles and should be used only where no other solution is possible.


The price to pay here is that some of these RAID arrays can cost in the million-dollar bracket and the configuration management of RAID levels and high availability features becomes almost esoteric . Allied with off-site data replication, the cost and configuration intricacies just go skyward. I once heard such a solution as being described as not a JBOD but a CSOB, a Complex Son Of a Bitch. You need to ask yourself, "What would it cost our organization to lose its data?" For some organizations, this cost is too much to bear; they wouldn't be able to operate without their data. The financial cost becomes almost inconsequential when considered against the possibility of a complete loss in business processing. We can answer the technical questions; the accountants and management need to answer the financial questions.



HP-UX CSE(c) Official Study Guide and Desk Reference
HP-UX CSE(c) Official Study Guide and Desk Reference
ISBN: N/A
EAN: N/A
Year: 2006
Pages: 434

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net