Sam Lightfoot created CASSANDRA-21572:
-----------------------------------------
Summary: Enable trickle_fsync by default
Key: CASSANDRA-21572
URL: https://issues.apache.org/jira/browse/CASSANDRA-21572
Project: Apache Cassandra
Issue Type: Improvement
Components: Local/SSTable
Reporter: Sam Lightfoot
Enable trickle_fsync by default in cassandra.yaml. It is already enabled in
cassandra_latest.yaml.
When enabled, SSTable writes are fsync'd at intervals controlled by
trickle_fsync_interval (default 10MiB) rather than a single fsync after the
SSTable is written. This bounds dirty page buildup and mitigates kernel
writeback storms, benefiting read latencies, see benchmarks in the comments on
[CASSANDRA-21487|https://issues.apache.org/jira/browse/CASSANDRA-21487].
Note trickle_fsync was silently ignored for compressed SSTables until
CASSANDRA-21487, so most operators have never actually run with it even when
enabled.
The historical caution was HDDs, where trickle fsync costs some sequential
throughput, but that cost falls on throttled background writes, whereas a
writeback storm falls on foreground reads. The property remains configurable.
dev@ discussion:
[https://lists.apache.org/thread/z9zhlh5f3orfbjmp9lcy8vhbk0obboq8.]
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
