d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Improve latency of synchronous 128KB writes.

Browse Source 
Before my ZIL space optimization few years ago 128KB writes were logged
as two 64KB+ records in two 128KB log blocks.  After that change it became
~124KB+/4KB+ in two 128KB log blocks to free space in the second block
for another record.  Unfortunately in case of 128KB only writes, when space
in the second block remained unused, that change increased write latency by
imbalancing checksum computation time between parallel threads.

This change introduces new 68KB log block size, used for both writes below
67KB and 128KB-sharp writes.  Writes of 68-127KB are still using one 128KB
block to not increase processing overhead.  Writes above 131KB are still
using full 128KB blocks, since possible saving there is small.  Mixed loads
will likely also fall back to previous 128KB, since code uses maximum of
the last 10 requested block sizes.

On a simple 128KB write test with queue depth of 1 this change demonstrates
~15-20% performance improvement.

MFC after:	2 weeks
Sponsored by:	iXsystems, Inc.
This commit is contained in:
Alexander Motin 2019-10-01 20:09:25 +00:00
parent 044477e294
commit e9f4580d92
1 changed files with 13 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zil.c
View File

@ -1351,11 +1351,17 @@ zil_lwb_write_open(zilog_t *zilog, lwb_t *lwb)
* aligned to 4KB) actually gets written. However, we can't always just
* allocate SPA_OLD_MAXBLOCKSIZE as the slog space could be exhausted.
*/
uint64_t zil_block_buckets[] = {
4096, /* non TX_WRITE */
8192+4096, /* data base */
32*1024 + 4096, /* NFS writes */
UINT64_MAX
struct {
uint64_t limit;
uint64_t blksz;
} zil_block_buckets[] = {
{ 4096, 4096 }, /* non TX_WRITE */
{ 8192 + 4096, 8192 + 4096 }, /* database */
{ 32768 + 4096, 32768 + 4096 }, /* NFS writes */
{ 65536 + 4096, 65536 + 4096 }, /* 64KB writes */
{ 131072, 131072 }, /* < 128KB writes */
{ 131072 + 4096, 65536 + 4096 }, /* 128KB writes */
{ UINT64_MAX, SPA_OLD_MAXBLOCKSIZE}, /* > 128KB writes */
};
/*
@ -1432,11 +1438,9 @@ zil_lwb_write_issue(zilog_t *zilog, lwb_t *lwb)
* pool log space.
*/
zil_blksz = zilog->zl_cur_used + sizeof (zil_chain_t);
for (i = 0; zil_blksz > zil_block_buckets[i]; i++)
for (i = 0; zil_blksz > zil_block_buckets[i].limit; i++)
continue;
zil_blksz = zil_block_buckets[i];
if (zil_blksz == UINT64_MAX)
zil_blksz = SPA_OLD_MAXBLOCKSIZE;
zil_blksz = zil_block_buckets[i].blksz;
zilog->zl_prev_blks[zilog->zl_prev_rotor] = zil_blksz;
for (i = 0; i < ZIL_PREV_BLKS; i++)
zil_blksz = MAX(zil_blksz, zilog->zl_prev_blks[i]);