ABSTRACT
Mean Time Between Failure (MTBF) is established before drives go into high-volume production, using a process called the Reliability Demonstration Test. MTBF is a useful litmus test to determine whether a new drive is reliable enough to ship, but it is preproduction use, not real-world. Use Service Level Agreements (SLAs) to control the expense to build and maintain. In the interest of fiscal responsibility, it is best to drive every application possible onto lower-tier SLAs. Remote replication and failover is expensive, so use it only when warranted. Local replication is a viable alternative. The tool should be focused on the task of affordably delivering storage performance, spending enough so the business task can complete smoothly, delivering on the SLA, and quickly identifying and resolving problems that get in the way of delivering the SLA, all the while avoiding overspending.
