Many of the standard ways that hard drives are used may not be appropriate for SSDs. Putting SSDs in a RAID 5 array, for instance, may make the RAID controller the bottleneck in the storage system, in addition to ensuring that all the drives in the array could fail at that same time, since writes are distributed evenly over all drives, and wear-leveling algorithms on the drives cause all cells to wear evenly as well, meaning that if the drives reach their limits on writes at the same time, all the drives in an array could fail at about the same time.
Since the mean time between failure for SSDs is higher than for hard drives, the best solution might be to implement a JBOD (just a bunch of disks) and use replication for fault tolerance. None of the vendors I spoke with currently offer monitoring of the remaining number of writes available on SSDs, though several say they are working on this.
Read more about data center in Network World's Data Center section.
This story, "Making the best use of SSDs" was originally published by Network World.