belliottsmith commented on code in PR #3157:
URL: https://github.com/apache/cassandra/pull/3157#discussion_r1522902019
##########
src/java/org/apache/cassandra/service/accord/AccordKeyspace.java:
##########
@@ -191,6 +203,15 @@ public class AccordKeyspace
private static final ClusteringIndexFilter FULL_PARTITION = new
ClusteringIndexSliceFilter(Slices.ALL, false);
+ //TODO (now, performance): should this be partitioner rather than TableId?
As of this patch distributed tables should only have 1 partitioner...
+ private static final ConcurrentMap<TableId,
AccordRoutingKeyByteSource.Serializer> TABLE_SERIALIZERS = new
ConcurrentHashMap<>();
Review Comment:
this is primarily a read-cache, so it might be nice to optimise for this and
not have the weight and cost of a concurrent hash map on read (e.g. CoW the
whole map but use an Object2Object open hash map).
We should also avoid infinite state growth under schema changes, so it
should either periodically be cleared, or have entries removed when tables are
dropped, or on update should clear any stale entries (if we're doing CoW this
is prob my favoured approach).
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
