Morgan Tocker (mtocker) wrote,
Morgan Tocker

Initial findings on EC2

Yesterday I reported that I was looking into EC2 to see how it performed as a MySQL Server. Well, I got around to running some benchmarks today. My home machine (AMD64 3000 (I think), Ubuntu 7.10, 7200RPM drive, 1G RAM) can handle almost 730 transactions/second on a basic sysbench test:

sysbench --test=oltp --mysql-table-engine=myisam --oltp-table-size=1000000 --mysql-socket=/var/run/mysqld/mysqld.sock  prepare
sysbench --num-threads=16 --max-requests=100000 --test=oltp --oltp-table-size=1000000 --mysql-socket=/var/run/mysqld/mysqld.sock --oltp-read-only run

What does EC2 (small instance) rate? 132.26 transactions/second.

I've got to say - I'm a little disappointed, since that's more than 5 times worse. I will test the other instance types over the coming days, and try and make my environment a little more controlled.

Databases are often disk i/o bound. Maybe they've considered this and improved it in the other available options? There's no details mentioning it though. Sigh. Correction: There is some mention.
Tags: mysql
  • Post a new comment


    default userpic

    Your reply will be screened

    Your IP address will be recorded 

    When you submit the form an invisible reCAPTCHA check will be performed.
    You must follow the Privacy Policy and Google Terms of use.
You have to us ethe large/x-large instances to get some real performance. If you're interested we can work together on some benchmarks. We have a CentOS5 server template for the large instances. Ubuntu not yet (need to build it). The larger instances not only have full cores (you get a time slice on the small instances) but also much better I/O.

Thorsten - CTO, RightScale
Hi Thorsten,

You are right - the large/x-large rate much better.

Small Instance = 132.26 transactions/per sec.
Large Instance = 319.18 transactions/per sec.

And this is inconclusive, since I just discovered that I may have used a pre-5.0.30 server (which suffers < a href="">bug 15815</a>).

Many thanks for the followup!
Morgan, I ran a bunch of benchmarks using InnoDB tables instead of the MyISAM tables and using a striped xfs filesystem. See for details. The large instances do scale up and particularly the write bandwidth is a lot better. Now I need to hunt for some comparables outside of EC2...
Thorsten -



July 22 2008, 20:02:47 UTC 10 years ago

n2Fash gfb07yvt9d6t94wbtx63bgq7d
onB8An hlgo8G3bdDa2Nx

Deleted comment

Webcam Chat Room (
Appreciate you sharing your findings Morgan. You might like to see some of the stuff we measured - including a fascinating finding on how CPU types affect the performance you get!