SQL IO Random vs Sequential

  • I'm trying to spin up a server in Amazon AWS that matches up the I/O specs in our current env by Provisioned IOPS. The benchmarking shows that AWS server has better random 8k block read and write, but worse on sequential 64k block read and write, as compared to my current env.

    My question is, should I match up my current env on Random or Sequential performance? Thanks!

  • I'd do both, but I'd also try to determine which is important in your current environment. What does your workload do more often?

  • Thanks for the advice. The thing is, if I'm trying to match both, it gets really expensive with Provisioned IOPS. So I'm trying to find a happy medium.

    We have both active reads and writes, probably 60% reads and 40% writes.

    Also, here's the comparison of an AWS drive I/O vs my current env - should I match on Reads or Writes? (I've found that there's no change in sequential I/O no matter the provisioned size)

    AWS:

    Both Random Reads/Writes 8k/block

    IOs/sec: 5000

    MBs/sec: 40

    Current Env:

    Random Reads 8k/block

    IOs/sec: 362.01

    MBs/sec: 2.82

    Random Writes 8k/block

    IOs/sec: 1640.18

    MBs/sec: 12.81

Viewing 3 posts - 1 through 2 (of 2 total)

You must be logged in to reply to this topic. Login to reply