SSD IOPS specs - can they actually be achieved?

I notice that fast SSDs now have an IOPS rating of about 100K/s (=0.01ms access time), however benchmark ratings are typically 10% of this - 0.1ms (=10K IOPS). Is the raw spec actually achievable with any real-world system? Would it require a highly optimised hardware controller, and perhaps without the overhead of a file system?

1 answer Last reply
More about iops specs achieved
  1. Doh - it's just command queuing! :)

    In this benchmark, they did indeed manage to get the IO rate up to the published spec, using queuing:

Ask a new question

Read More

SSD Benchmark Storage