[
https://issues.apache.org/jira/browse/CASSANDRA-5019?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13809322#comment-13809322
]
Benedict commented on CASSANDRA-5019:
-------------------------------------
Ah, my mistake, I was looking at the CQL query processor. Either way, not to
push the issue, but it's not *absolutely* necessary; an
IncrementalRowSerializer (with an IncrementalColumnFamilySerializer) or similar
could be defined which collects the atoms and serializes them immediately. Only
needs to be used as an optimisation on the common code paths to make a big
difference, and the remainder could simply invoke a wrapper method that
constructs the Row as usual. The Row/ColumnFamily serializers could trivially
be switched to calling these incremental serializers, so no code duplication,
hot code paths can be updated without *tremendous* effort, and infrequent code
paths take an imperceptible performance hit (given the context will be stack
allocated it is unlikely to be measureable).
> Still too much object allocation on reads
> -----------------------------------------
>
> Key: CASSANDRA-5019
> URL: https://issues.apache.org/jira/browse/CASSANDRA-5019
> Project: Cassandra
> Issue Type: Improvement
> Components: Core
> Reporter: Jonathan Ellis
> Assignee: Yuki Morishita
> Labels: performance
> Fix For: 2.1
>
>
> ArrayBackedSortedColumns was a step in the right direction but it's still
> relatively heavyweight thanks to allocating individual Columns.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
