<html><body><div style="color:#000; background-color:#fff; font-family:arial, helvetica, sans-serif;font-size:12pt"><div><span>Joe,</span></div><div><span>Thanks for spending time with aio_stress(much apperciated) and the information. A week back I ran the tests on flash device and saw lower numbers. However all my runs since last week have been with ramdisk only. I run the following command to make sure</span> <span>ramdisk has backing store.</span></div><br><div><span>[root@lab-dm-cn4 ~]# dd if=/dev/zero of=/dev/ram bs=512 oflag=direct count=`</span><span>blockdev --getsize /dev/ram`</span><span><br>8388608+0 records in<br>8388608+0 records out<br>4294967296 bytes (4.3 GB) copied, 19.2371 s, 223 MB/s<br></span></div><div><span><br></span></div>One interesting thing I notice is raw ramdisk has 9000+MB/s throughput in my tests and 5440MB/s throughput in your tests. I remember you mentioning that you are running tests in a VM with 4G memory,
while I run the rests on a standalone server with 16 quad-core processors and 12G or ram (out of that 12G, I carve out 4G ramdisk). Attached are the cpu and memory information. Could that be causing an issue?<br><br>Thanks,<br>Jagan.<br><div><br></div> <div style="font-family: arial,helvetica,sans-serif; font-size: 12pt;"> <div style="font-family: times new roman,new york,times,serif; font-size: 12pt;"> <div dir="ltr"> <font face="Arial" size="2"> <hr size="1"> <b><span style="font-weight: bold;">From:</span></b> Joe Thornber <thornber@redhat.com><br> <b><span style="font-weight: bold;">To:</span></b> Jagan Reddy <gjmsreddy@yahoo.com>; device-mapper development <dm-devel@redhat.com> <br> <b><span style="font-weight: bold;">Sent:</span></b> Friday, January 20, 2012 10:07 AM<br> <b><span style="font-weight: bold;">Subject:</span></b> Re: [dm-devel] dm-thin vs lvm performance<br> </font> </div> <br>
Jagan,<br><br>On Fri, Jan 20, 2012 at 05:33:34PM +0000, Joe Thornber wrote:<br>> Back to testing ...<br><br>Here are my aio-stress results for various devices running on top of a<br>ramdisk:<br><br>| Device stack | M/s | <br>+------------------------+------+ <br>| raw ramdisk | 5440 |<br>| linear | 5431 |<br>| 2 stacked linear | 5304 |<br>| pool device | 5351 |<br>| linear stacked on pool | 5243 |<br>| thin | 5375 |<br><br>I also tried the thin test after disabling the block locking in<br>dm-block-manager.c, and using a 128 entry cache local to the thin<br>device. Neither made any difference to the 'thin' result.<br><br>Things for you to check:<br><br>i) Are you benchmarking
with a ramdisk, or your flash device? In both<br> cases I think you need to make sure the device has allocated backing<br> store before you run any tests. Remember that Discards may well<br> removing this backing.<br><br>ii) You claim that io is getting deferred. This will happen when you do<br> the initial wipe of the thin device to force provisioning. But for<br> the aio-stress tests you should get nothing deferred (I instrumented<br> to confirm this).<br><br><br>- Joe<br><br><br> </div> </div> </div></body></html>