acelyc111 commented on a change in pull request #5597:
URL: https://github.com/apache/incubator-doris/pull/5597#discussion_r608305362
##########
File path: be/src/common/config.h
##########
@@ -516,10 +514,23 @@ CONF_Int32(flush_thread_num_per_store, "2");
// config for tablet meta checkpoint
CONF_mInt32(tablet_meta_checkpoint_min_new_rowsets_num, "10");
CONF_mInt32(tablet_meta_checkpoint_min_interval_secs, "600");
+// Thread count of RocksDB uses for background flush and compaction, -1 means
the number of cores.
+CONF_mInt32(rocksdb_thread_count, "-1");
+CONF_mInt32(rocksdb_block_cache_mb, "1024");
+CONF_Validator(rocksdb_block_cache_mb, [](int value) -> bool {
+ return value > 0;
+});
+CONF_String(rocksdb_compression_type, "LZ4");
Review comment:
> Use options.compression to specify the compression to use. By default
it is Snappy. We believe LZ4 is almost always better than Snappy. We leave
Snappy as default to avoid unexpected compatibility problems to previous users.
https://github.com/facebook/rocksdb/wiki/Compression
We encountered a problem that snappy decompress cost too much time in
product environment.
Of course, the root cause is that the rocksdb value is too large, some
tablet meta which have thounds of rowsets may reach hundreds of MB. But I think
we can improve it by use LZ4.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
