Prelim Memoright SSD Tests

I picked up what some claim is the fastest of the current SSD drives this week, the Memoright GT.  From articles around the net I have seen performance speeds substantially faster then the mtron drive I tested earlier…. so I took the plunge.  The first results?  luke warm.  The random performance ( databases like mysql are all about random performance )  of the drive is better in mixed read/write tests over the mtron, but the mtron blew it away in random read performance.  This are are my first pass tests, so I may have something wrong …  so take them with a grain of salt:

Req Per Second
rnd read/write 1 Raptor 1 Mtron 1 memoright
5000/5000 172 200 248.18
6670/3330 164 282 340.97
7500/2500 159 388 433.93
8000/2000 165 516 456.34
8333/1777 176 518 563.09
10000/0 161 5263 1364.02
0/10000 200 100 140.59

The split tests were higher in all but 1, but the random read alone was 5x slower! more later.

This entry was posted in hardware, linux, mysql. Bookmark the permalink.

2 Responses to Prelim Memoright SSD Tests

  1. 2fast4ya says:

    This is very interesting test/review!

    What is the hardware setup that you have used for the tests?

    The 100% read results seem a bit odd to me as well as the 8000/2000.

    What size are the requests?

    Have you seen this?

    http://www.memoright-rep.eu/pdf/EU%20MR25-2-064S%20vs%20MTRON%20PRO%207500%20IOmeter.pdf

    Regards,
    2fast4ya

  2. matt says:

    I have seen the results. I was able to get a peak of 4K iops @ 100 reads over the weekend with a little tweaking. I realized the posted tests were using a better controller card then I used for the mtron, so I switched back to the same controller as the mtron. But what is odd is after several heavy read/write tests the next read only test sometimes suffers ( <1500 iops -vs- 3500-4000 iops ). I wonder if they are doing some voodoo behind the scenes to speed up writes ( i.e. writing them sequentially and coming back to mark the random block later ala the mft stuff I tested earlier ). Whats more troublesome is the Latency of the Memoright vs the Mtron. In dbt2 tests I see 8-12ms wait times for disk, in the same test with the mtron I was < 2ms. Additionally when I publish my final results you will see the dbt2 test results were in some cases 2x lower then the mtron.

    My tests were done with 16K io, which is what innodb uses internally. All these tests are done on a 4 core machine, with 6gb of memory. filesystem built with a 4MB journal size, 4K block, and 4K fragment size.