On Wed, 8 Apr 2015, Jason Matthews wrote:
Do you have any notion what if any performance penalty there is staying with
512 byte sectors?
I have no idea what the penalty with your SSDs would be, but the
penalty should only be when writing.
SSDs need to erase a block before they can write. W
Do you have any notion what if any performance penalty there is staying
with 512 byte sectors?
Given the cost of high quality SSDs I am not inclined to move to larger
sector sizes unless there is a real pay off. Right now, I dont see it.
j.
On 4/8/2015 5:41 PM, Matthew Ahrens wrote:
On We
cross posting to zfs
The data seems to indicate that something strange is happening between
the compression algorithms and sector sizes > 512 bytes. ZFS reports
virtually the same compression ratios for either ashift=9 or ashift=12
zpools, but the number of data blocks consumed varies drasti
On 08/04/2015 15:32, Jason Matthews wrote:
I am going to try again with 128k. Should know something in a few hours.
Here are my test results:
Data set 1
zfs settings | block size
compress recordsize | 512 4k ratio
--
off 12
I am going to try again with 128k. Should know something in a few hours.
Sent from my iPhone
> On Apr 8, 2015, at 2:31 AM, James wrote:
>
>> On 08/04/2015 08:21, Jason Matthews wrote:
>>
>> Are there inefficiencies that creep into LZ4 with larger block sizes
>> some how?
>
> Are you using z
Yes.
Sent from my iPhone
> On Apr 8, 2015, at 2:31 AM, James wrote:
>
>> On 08/04/2015 08:21, Jason Matthews wrote:
>>
>> Are there inefficiencies that creep into LZ4 with larger block sizes
>> some how?
>
> Are you using zfs recordsize=8k? (As often recommended with Postgresql.)
>
>
>
On 08/04/2015 08:21, Jason Matthews wrote:
Are there inefficiencies that creep into LZ4 with larger block sizes
some how?
Are you using zfs recordsize=8k? (As often recommended with Postgresql.)
James.
___
openindiana-discuss mailing list
openin
> On 4/7/2015 1:13 PM, Jason Matthews wrote:
> >
> > I wonder what 4k write sizes look like. Let's find out.
>
> The results are in. 4k sectors consume 302G for this data set.
>
> 512b 202GB
> 4KB302GB
> 8KB474GB
Wow. That is quite... shocking. Thanks for measuring this.
We try to stan
On 4/7/2015 2:36 PM, Jason Matthews wrote:
The results are in. 4k sectors consume 302G for this data set.
512b 202GB
4KB302GB
8KB474GB
It seems that George Wilsons modification to the sd driver to let you
nail the sector size only works in one direction. You can only specify a
s
