[
https://issues.apache.org/jira/browse/HBASE-21098?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16603848#comment-16603848
]
Mingliang Liu commented on HBASE-21098:
---------------------------------------
{quote}
For TestSnapshotDFSTemporaryDirectory, through my debugging, it showed that the
file system of both the working and root directory were DFS rather than the
local filesystem
{quote}
You are right. I debugged and checked the working directory FS, it is correctly
populated as DFS for the unit test. I was wrongly thinking the precondition
should have failed. Thanks for clarification.
{quote}
Do you think it's not worth making the logic more complicated for the occasions
that the rename would be faster?
{quote}
If the complexity is affordable, I think we should make it faster. Rename will
more preferred to the copy (and delete?). One possible approach is to 1)
compare the FileSystem _scheme_, _authority_, and _ugi_ to see if it's the
"same" filesystem backend instead of {{equals()}} 2) failover to copy if rename
fails for any reason (as your code is doing now)
Thanks,
> Improve Snapshot Performance with Temporary Snapshot Directory when rootDir
> on S3
> ---------------------------------------------------------------------------------
>
> Key: HBASE-21098
> URL: https://issues.apache.org/jira/browse/HBASE-21098
> Project: HBase
> Issue Type: Improvement
> Affects Versions: 3.0.0, 1.4.8, 2.1.1
> Reporter: Tyler Mi
> Priority: Major
> Attachments: HBASE-21098.master.001.patch,
> HBASE-21098.master.002.patch, HBASE-21098.master.003.patch,
> HBASE-21098.master.004.patch, HBASE-21098.master.005.patch,
> HBASE-21098.master.006.patch, HBASE-21098.master.007.patch,
> HBASE-21098.master.008.patch, HBASE-21098.master.009.patch,
> HBASE-21098.master.010.patch
>
>
> When using Apache HBase, the snapshot feature can be used to make a point in
> time recovery. To do this, HBase creates a manifest of all the files in all
> of the Regions so that those files can be referenced again when a user
> restores a snapshot. With HBase's S3 storage mode, developers can store their
> data off-cluster on Amazon S3. However, utilizing S3 as a file system is
> inefficient in some operations, namely renames. Most Hadoop ecosystem
> applications use an atomic rename as a method of committing data. However,
> with S3, a rename is a separate copy and then a delete of every file which is
> no longer atomic and, in fact, quite costly. In addition, puts and deletes on
> S3 have latency issues that traditional filesystems do not encounter when
> manipulating the region snapshots to consolidate into a single manifest. When
> HBase on S3 users have a significant amount of regions, puts, deletes, and
> renames (the final commit stage of the snapshot) become the bottleneck
> causing snapshots to take many minutes or even hours to complete.
> The purpose of this patch is to increase the overall performance of snapshots
> while utilizing HBase on S3 through the use of a temporary directory for the
> snapshots that exists on a traditional filesystem like HDFS to circumvent the
> bottlenecks.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
