[ltp] Slightly OT: Filesystem performance on SSD vs HDD: test data request

[Theodore Tso]

On Mon, Aug 17, 2009 at 08:22:42PM +0100, Richard Neill wrote:
> Dear All,
>
> I've been looking into filesystem performance for postgresql on various  
> configurations. I've found out quite a lot. I'm still curious about how  
> it works on a really good SSD; does anyone have an Intel X-25 that they  
> would be willing to run a simple test with?

I did the test on a T400 (Core 2.53 GHz), using LVM and mke2fs
parameters of "mke2fs -b 4096 -t ext[234] /dev/<PV>/test 131072".
This means that I was using a half gigabyte test partition, freshly
created, on an otherwise idle system.  (If you have any other I/O
writing to the same disk, or especially the same filesystem,
fdatasync's performance numbers will be very, very different.)  I am
running a 2.6.31-rc5 kernel.

On the 5400rpm HDD in the Ultrabay slot:

hdparm -t	66 MB/sec
ext2		1.4s
ext3		2.2s
ext4/barrier=0	2.2s
ext4/barrier=1	29.3s

On the Intel X25-M 80GB SSD as the primary disk:

hdparm -t	157 MB/sec
ext2		0.23s
ext3		0.46s
ext4/barrier=0	0.46s
ext4/barrier=1	2.90s

Note that ext2/ext3 defaults to barrier=0, while ext4 uses barrier=1
as the default.  So I did the test both ways in order to have a fair
apples-to-apples comparison.

						- Ted