> On Aug 18, 2016, at 1:58 PM, Jose Luis Rodriguez Garcia <[email protected]>
> wrote:
>
> On Thu, Aug 18, 2016 at 7:24 PM, Eduardo Horvath <[email protected]> wrote:
>>
>> LFS writes the metadata at the same time, in the same place as the data.
>> No synchronous writes necessary.
>
> As I understand LFS needs to do synchronous writes when there is
> metadata operations (directories)/fsync operations involved. Instead
> of writting a full segment (1 MB per default), it writes a "small
> segment". It kills performance in RAID 5/6, because the write isn't a
> full stripe: you have to read all the disks, for calculate the new
> parity 1 write on raid of x disks= x reads + 2 writes (data + parity).
That's not correct.
You're right that partial stripe writes are more expensive, but they are not
*that* expensive. For RAID 5, the cost is two read and two writes: read/write
of the data disk, and read/write of the parity disk. For RAID 6 is it three
reads and writes (read/write on both of the P and Q disks).
paul
