Apache
/
hbase
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

HBASE-20337 Update the doc on how to setup shortcircuit reads; its stale

This commit is contained in:
Michael Stack 2018-04-03 10:27:38 -07:00
parent 0c0fe05bc4
commit d60decd959
1 changed files with 22 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
src/main/asciidoc/_chapters/schema_design.adoc
View File

@ -1148,16 +1148,36 @@ Detect regionserver failure as fast as reasonable. Set the following parameters:
- `dfs.namenode.avoid.read.stale.datanode = true`
- `dfs.namenode.avoid.write.stale.datanode = true`
[[shortcircuit.reads]]
=== Optimize on the Server Side for Low Latency
Skip the network for local blocks when the RegionServer goes to read from HDFS by exploiting HDFS's
link:https://hadoop.apache.org/docs/stable/hadoop-project-dist/hadoop-hdfs/ShortCircuitLocalReads.html[Short-Circuit Local Reads] facility.
Note how setup must be done both at the datanode and on the dfsclient ends of the conneciton -- i.e. at the RegionServer
and how both ends need to have loaded the hadoop native `.so` library.
After configuring your hadoop setting _dfs.client.read.shortcircuit_ to _true_ and configuring
the _dfs.domain.socket.path_ path for the datanode and dfsclient to share and restarting, next configure
the regionserver/dfsclient side.
* Skip the network for local blocks. In `hbase-site.xml`, set the following parameters:
* In `hbase-site.xml`, set the following parameters:
- `dfs.client.read.shortcircuit = true`
- `dfs.client.read.shortcircuit.buffer.size = 131072` (Important to avoid OOME)
- `dfs.client.read.shortcircuit.skip.checksum = true` so we don't double checksum (HBase does its own checksumming to save on i/os. See <<hbase.regionserver.checksum.verify.performance>> for more on this.
- `dfs.domain.socket.path` to match what was set for the datanodes.
- `dfs.client.read.shortcircuit.buffer.size = 131072` Important to avoid OOME -- hbase has a default it uses if unset, see `hbase.dfs.client.read.shortcircuit.buffer.size`; its default is 131072.
* Ensure data locality. In `hbase-site.xml`, set `hbase.hstore.min.locality.to.skip.major.compact = 0.7` (Meaning that 0.7 \<= n \<= 1)
* Make sure DataNodes have enough handlers for block transfers. In `hdfs-site.xml`, set the following parameters:
- `dfs.datanode.max.xcievers >= 8192`
- `dfs.datanode.handler.count =` number of spindles
Check the RegionServer logs after restart. You should only see complaint if misconfiguration.
Otherwise, shortcircuit read operates quietly in background. It does not provide metrics so
no optics on how effective it is but read latencies should show a marked improvement, especially if
good data locality, lots of random reads, and dataset is larger than available cache.
For more on short-circuit reads, see Colin's old blog on rollout,
link:http://blog.cloudera.com/blog/2013/08/how-improved-short-circuit-local-reads-bring-better-performance-and-security-to-hadoop/[How Improved Short-Circuit Local Reads Bring Better Performance and Security to Hadoop].
The link:https://issues.apache.org/jira/browse/HDFS-347[HDFS-347] issue also makes for an
interesting read showing the HDFS community at its best (caveat a few comments).
=== JVM Tuning
==== Tune JVM GC for low collection latencies