Re: large object write performance

Op 08-10-15 om 13:21 schreef Graeme B. Bell:
>>> First the database was on a partition where compression was enabled, I changed it to an uncompressed one to see if
itmakes a difference thinking maybe the cpu couldn't handle the load. 
>> It made little difference in my case.
>>
>> My regular gmirror partition seems faster:
>> dd bs=8k count=25600 if=/dev/zero of=./test
>> 25600+0 records in
>> 25600+0 records out
>> 209715200 bytes transferred in 1.513112 secs (138598612 bytes/sec)
>>
>> the zfs compressed partition also goes faster:
>> dd bs=8k count=25600 if=/dev/zero of=./test
>> 25600+0 records in
>> 25600+0 records out
>> 209715200 bytes transferred in 0.979065 secs (214199479 bytes/sec)
>> but this one didn't really go that fast in my test (maybe 10%)
>
> Please can you run iozone and look for low random write performance with small blocks?  (4k)
> http://www.slashroot.in/linux-file-system-read-write-performance-test
Like this ?

gmirror (iozone -s 4 -a /dev/mirror/gm0s1e) = 806376 (faster drives)
zfs uncompressed (iozone -s 4 -a /datapool/data) = 650136
zfs compressed (iozone -s 4 -a /datapool/data) = 676345

> Also please can you CC to the list with your replies to my on-list emails?
>
> Graeme Bell
>
>