Server Fault is a question and answer site for system and network administrators. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

How fast a disk can perform?

Does 700k writing operations per second a reasonable upper bound?

Each operation is a random write with 1024 Bytes as block size .

share|improve this question
What's "normal"? How big is an "operation"? A single block write? Something bigger? A seek? A seek across the entire surface? A seek from middle to outer edge? – S.Lott May 18 '09 at 16:48
This question needs much, much, much more detail. In addition to S.Lott's questions, what kind of hard drive? How fast? How large? Are you observing 700k writing ops/second, or trying to get there? What is the purpose of this question? – sh-beta May 18 '09 at 17:18

After doing some quick googling and checking wikipedia, seems like a typical 7200RPM SATA drive is capable of somewhere between 50-100 I/O operations per second.

So 700k seems very, very high.

share|improve this answer

The actual value depends on a number of things, buffer size, sequential read/write speeds, random read/write speeds, just depends on what type of operations you're running.

700k should be an acceptable upper bound for most applications (I would say most hard drives can do more than 10MB/s for both random and sequential read/writes).

share|improve this answer
You've obviously never actually benchmarked any of this. 700k sectors per second is 341MB/s, which is in "yeah, good luck" territory for sequential writes. 700,000 random operations per second is fantasy land. – womble May 18 '09 at 22:40
I think you mixed 700k (as in 700 thousand) operations with 700KB (as in Kilobytes)... In that case, 700KB/sec sounds perfectly feasible for disks that handle 10MB/sec :) – Rafa Apr 17 '12 at 8:46

Presumably you are talking about bandwidth when you say 700k per second, not IOPS (I/O per second)?

Either way, when you're doing 1K-size random writes across the entire disk you are probably being buffered by the operating system. Random writes like that will get cached and written out to the disk in a more sensible manner.

It's also important to note - when you say disk, are you talking about:

  • A block device? (i.e. RAID, SAN disk, etc.)
  • A SATA drive?
  • A SAS drive?
  • A Fiber Channel drive?

Each of these options will get you a different answer.

If you want to really test your disk in a sensible manner, I recommend checking out bonnie++.

share|improve this answer

Depends how random/sequential the writes are, if they're sequential any you can increase your block size you shouldn't have a problem at all. If they're random/reasonably-random and you stick with 1K blocks you'll need an array.

share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.