librbd: tweaks to improve throughput for journaled IO

Increase the per-object append flush bytes to 1MiB to mimic the 16MiB
in-memory writeback cache. Finally, default enable the new journal
writethrough until flush option.

Performance results comparing no journal, pre-change, and post-change:

	no journal	original	modified
4K	40310.84	7223.73		27427.27
32K	21530.37	3256.39		10284.05
256K	3159.67		564.73		1627.16
4M	223.03		48.03		91.91

Fixes: http://tracker.ceph.com/issues/40072
Signed-off-by: Jason Dillaman <dillaman@redhat.com>

diff --git a/src/common/options.cc b/src/common/options.cc
index 167bb45044ec4c7ed86616817bc819a55f79a7d9..f8f32119878c6713285e819f972efb3f97786b91 100644 (file)
--- a/src/common/options.cc
+++ b/src/common/options.cc
@@ -7305,7 +7305,7 @@ static std::vector<Option> get_rbd_options() {
     .set_description("commit time interval, seconds"),
 
     Option("rbd_journal_object_writethrough_until_flush", Option::TYPE_BOOL, Option::LEVEL_ADVANCED)
-    .set_default(false)
+    .set_default(true)
     .set_description("when enabled, the rbd_journal_object_flush* configuration "
                      "options are ignored until the first flush so that batched "
                      "journal IO is known to be safe for consistency"),
@@ -7315,7 +7315,7 @@ static std::vector<Option> get_rbd_options() {
     .set_description("maximum number of pending commits per journal object"),
 
     Option("rbd_journal_object_flush_bytes", Option::TYPE_SIZE, Option::LEVEL_ADVANCED)
-    .set_default(0)
+    .set_default(1_M)
     .set_description("maximum number of pending bytes per journal object"),
 
     Option("rbd_journal_object_flush_age", Option::TYPE_FLOAT, Option::LEVEL_ADVANCED)