[ https://issues.apache.org/jira/browse/HBASE-5074?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13215115#comment-13215115 ]
Phabricator commented on HBASE-5074: ------------------------------------ mbautin has commented on the revision "[jira] [HBASE-5074] Support checksums in HBase block cache". @dhruba: going through the diff once again. Since you've updated the revision, submitting existing comments against the previous version, and continuing with the new version. INLINE COMMENTS src/main/java/org/apache/hadoop/hbase/io/hfile/ChecksumUtil.java:131 Misspelling: "Minimun" -> Minimum src/main/java/org/apache/hadoop/hbase/fs/HFileSystem.java:44-45 Can these two be made final too? src/main/java/org/apache/hadoop/hbase/io/hfile/ChecksumUtil.java:145 s/chuck/chunk/ src/main/java/org/apache/hadoop/hbase/io/hfile/AbstractHFileReader.java:48 Fix javadoc: do do -> do src/main/java/org/apache/hadoop/hbase/io/hfile/ChecksumUtil.java:38 Make this final, rename to DUMMY_VALUE, because this is a constant, and make the length a factor of 16 to take advantage of alignment. src/main/java/org/apache/hadoop/hbase/io/hfile/FixedFileTrailer.java:532 s/manor/major/ src/main/java/org/apache/hadoop/hbase/io/hfile/HFile.java:157 This comment is misleading. This is not something that defaults to the 16 K, but the default value itself. I think this should say something about how a non-default value is specified. src/main/java/org/apache/hadoop/hbase/io/hfile/HFile.java:265-271 The additional constructor should not be needed when https://issues.apache.org/jira/browse/HBASE-5442 goes in. src/main/java/org/apache/hadoop/hbase/io/hfile/HFile.java:409 Is it possible to obtain the filesystem from the input stream rather than pass it as an additional parameter? Or is the underlying filesystem of the input stream a regular one, as opposed to an HFileSystem? REVISION DETAIL https://reviews.facebook.net/D1521 > support checksums in HBase block cache > -------------------------------------- > > Key: HBASE-5074 > URL: https://issues.apache.org/jira/browse/HBASE-5074 > Project: HBase > Issue Type: Improvement > Components: regionserver > Reporter: dhruba borthakur > Assignee: dhruba borthakur > Attachments: D1521.1.patch, D1521.1.patch, D1521.2.patch, > D1521.2.patch, D1521.3.patch, D1521.3.patch, D1521.4.patch, D1521.4.patch, > D1521.5.patch, D1521.5.patch, D1521.6.patch, D1521.6.patch, D1521.7.patch, > D1521.7.patch, D1521.8.patch, D1521.8.patch, D1521.9.patch, D1521.9.patch > > > The current implementation of HDFS stores the data in one block file and the > metadata(checksum) in another block file. This means that every read into the > HBase block cache actually consumes two disk iops, one to the datafile and > one to the checksum file. This is a major problem for scaling HBase, because > HBase is usually bottlenecked on the number of random disk iops that the > storage-hardware offers. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira