[
https://issues.apache.org/jira/browse/CASSANDRA-14556?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16559633#comment-16559633
]
ASF GitHub Bot commented on CASSANDRA-14556:
--------------------------------------------
Github user iamaleksey commented on a diff in the pull request:
https://github.com/apache/cassandra/pull/239#discussion_r205747034
--- Diff:
src/java/org/apache/cassandra/db/streaming/CassandraOutgoingFile.java ---
@@ -114,11 +165,77 @@ public void write(StreamSession session,
DataOutputStreamPlus out, int version)
CassandraStreamHeader.serializer.serialize(header, out, version);
out.flush();
- CassandraStreamWriter writer = header.compressionInfo == null ?
- new CassandraStreamWriter(sstable,
header.sections, session) :
- new
CompressedCassandraStreamWriter(sstable, header.sections,
-
header.compressionInfo, session);
- writer.write(out);
+ if (shouldStreamEntireSSTable() && out instanceof
ByteBufDataOutputStreamPlus)
+ {
+ CassandraBlockStreamWriter writer = new
CassandraBlockStreamWriter(sstable, session, manifest);
+ writer.write((ByteBufDataOutputStreamPlus) out);
+ }
+ else
+ {
+ CassandraStreamWriter writer = (header.compressionInfo ==
null) ?
+ new CassandraStreamWriter(sstable, header.sections,
session) :
+ new CompressedCassandraStreamWriter(sstable,
header.sections,
+
header.compressionInfo, session);
+ writer.write(out);
+ }
+ }
+
+ @VisibleForTesting
+ public boolean shouldStreamEntireSSTable()
+ {
+ // don't stream if full sstable transfers are disabled or legacy
counter shards are present
+ if (!isZeroCopySSTableTransfersEnabled ||
ref.get().getSSTableMetadata().hasLegacyCounterShards)
+ return false;
+
+ ColumnFamilyStore cfs =
ColumnFamilyStore.getIfExists(getTableId());
+
+ if (cfs == null)
+ return false;
+
+ AbstractCompactionStrategy compactionStrategy =
cfs.getCompactionStrategyManager()
+
.getCompactionStrategyFor(ref.get());
+
+ if (compactionStrategy instanceof LeveledCompactionStrategy)
+ return contained(ranges, ref.get());
+
+ if (compactionStrategy instanceof SizeTieredCompactionStrategy)
+ {
+ return (ranges != null
+ && ranges.size() == 1
+ && ranges.get(0)
+ .contains(new
Range<>(ref.get().first.getToken(),
--- End diff --
Should be Bounds for the sstable, since range is (first, last] and sstable
contains [first, last] (thanks @krummas). But on a higher level note, maybe
just stick to LCS only for now, and enable STCS after we commit
CASSANDRA-14568? We are going to do that in 4.0, so this temporary special case
won't make it to a release anyway ¯\_(ツ)_/¯
> Optimize streaming path in Cassandra
> ------------------------------------
>
> Key: CASSANDRA-14556
> URL: https://issues.apache.org/jira/browse/CASSANDRA-14556
> Project: Cassandra
> Issue Type: Improvement
> Components: Streaming and Messaging
> Reporter: Dinesh Joshi
> Assignee: Dinesh Joshi
> Priority: Major
> Labels: Performance
> Fix For: 4.x
>
>
> During streaming, Cassandra reifies the sstables into objects. This creates
> unnecessary garbage and slows down the whole streaming process as some
> sstables can be transferred as a whole file rather than individual
> partitions. The objective of the ticket is to detect when a whole sstable can
> be transferred and skip the object reification. We can also use a zero-copy
> path to avoid bringing data into user-space on both sending and receiving
> side.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
