GitHub user lianhuiwang opened a pull request:
https://github.com/apache/spark/pull/10277
[SPARK-4621][shuffle]Shuffle index can cached for SortShuffleManager in
ExternalShuffle in order to reduce indexFile's io
In ExternalShuffle shuffle index can cached for SortShuffleManager at
first. Next time another reduce's fetching can read index from cache. Because
reduce tasks fetch map output from ExternalShuffle at same time, shuffle index
cache is effective.
In my test that it has 1000 partitions and 100 map tasks, it spends 3882ms
for no cache index and it spends 608ms for cache index. So it is six times
faster than index file.
@rxin @JoshRosen @andrewor14
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/lianhuiwang/spark SPARK-4621
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/spark/pull/10277.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #10277
----
commit da3e330411c602f1d1816333cc7574083f5737f0
Author: Lianhui Wang <[email protected]>
Date: 2015-12-12T14:42:01Z
init commit
----
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
