[ https://issues.apache.org/jira/browse/HDFS-6865?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Todd Lipcon updated HDFS-6865: ------------------------------ Status: Patch Available (was: Open) marking patch available since the JIRA this depends on is now committed. James, mind copying the brief description of the changes in this patch here to help guide reviewers? Have you done any benchmark to see whether there is a negative performance impact for any workloads? Might be interesting to compare the following before/after patch: - Directly instantiate a stream into HDFS, with default buffer size (using fs.open). Perform writes of small data (eg using a single-byte write((byte)123) type call in a loop) - Same as above, but wrap the stream in a BufferedOutputStream with a large buffer - Above two cases with larger writes (eg a few hundred bytes) Running something like the TestHLogPerformance from HBase would also be good to verify for no regressions > Byte array native checksumming on client side (HDFS changes) > ------------------------------------------------------------ > > Key: HDFS-6865 > URL: https://issues.apache.org/jira/browse/HDFS-6865 > Project: Hadoop HDFS > Issue Type: Sub-task > Components: hdfs-client, performance > Reporter: James Thomas > Assignee: James Thomas > Attachments: HDFS-6865.patch > > -- This message was sent by Atlassian JIRA (v6.2#6252)