2100 and SSD's

engine411 wrote:
Another way of asking it is, with the current hard drives, if the NAS is completely maxed out in throughput while moving data (not writing, but reading), which component is maxed out and holding it from allowing any more throughput? The NICs or the drives?

For sequential I/O (large files) the answer probably depends on the details of the RAID mode and the traditional disks you are comparing the SSDs with.

For small files, directory browsing/searching, database updates (other non-sequential access), then SSDs should make a huge difference since they have no latency when seeking. That would be your IOPS / throughput distinction. If you want to maximize IOPS, then SSDs are what you want (assuming you can afford them).

Of course SSDs would eliminate any possibility of the drives being the bottleneck (no matter what the I/O load mix looks like).

BTW, I am not convinced there is much benefit to configuring SSDs into a RAID array - I don't think their failure modes are well matched to raid protection. So you might consider jbod if you go with SSDs. The main downside is that you need to balance the capacity across the disks manually. As always, if you care about your data, you need a backup plan.

BTW2, though most studies show that with traditional drives, reliability of enterprise and consumer models are quite similar, that is not the case with SSD. SLC drives [generally targeted for enterprise] are faster and more reliable than MLC [targeted for consumer]. The number of write cycles is much higher with SLC.