Fully Automated RAID Level Selection Techniques for Disk Arrays Page 3So, the short of it is that there is no single best choice of RAID
- Selecting the type and number of arrays,
- Selecting the size and RAID level for each LU in each disk array, and
- Placing stores on the resulting LUs.
In other words, the administrator's goals are operational in nature (such as minimum cost or maximum reliability for a given cost) while satisfying the performance requirements of client applications. This is clearly a very difficult task, so manual approaches apply rules of thumb and gross over-provisioning to simplify the problem (i.e. stripe each database table over as many RAID 1/0 LUs as possible). Unfortunately, the resulting configurations can cost as much as two to three times more than necessary. This is especially relevant when the cost of a large SAN is easily measured in millions of dollars or when the cost of the SAN represents more than half the total system hardware cost. Perhaps even more important is the uncertainty that surrounds a manually-designed system -- how well will it meet its performance and availability goals?
Therefore, with the preceding in mind, it is believed that the automated design and configuration of SANs can overcome these limitations, as they can consider a wider range of workload interactions and can explore a great deal more of the search space than any manual method. To do so, though, these automated methods need to be able to make RAID-level selection decisions.