Pankaj Gupta 2012-11-09, 01:43
-Re: Throughput of HDFSSink
Brock Noland 2012-11-09, 16:10
For performance testing I
highly recommend org.apache.flume.source.StressSource
Perhaps try that?
On Thu, Nov 8, 2012 at 7:43 PM, Pankaj Gupta <[EMAIL PROTECTED]> wrote:
> What is the throughput I can expect when writing to the HDFS Sink. Here is
> the flume config I'm using:
> # in this case called 'agent'
> # Define a memory channel called ch1 on agent1
> agent1.channels.ch1.type = memory
> # Define an exec source called exec-source1 on agent1 and tell it
> # to bind to 0.0.0.0:41414. Connect it to channel ch1.
> agent1.sources.exec-source1.channels = ch1
> agent1.sources.exec-source1.type = exec
> agent1.sources.exec-source1.restart = true
> agent1.sources.exec-source1.batchSize = 100
> agent1.sources.exec-source1.command = /home/ubuntu/flume/linesource.sh
> # Define a logger sink that simply logs all events it receives
> # and connect it to the other end of the same channel.
> agent1.sinks.hdfs-sink1.channel = ch1
> agent1.sinks.hdfs-sink1.type = hdfs
> agent1.sinks.hdfs-sink1.hdfs.path > hdfs://ip-10-000-000-000.ec2.internal/user/ubuntu/event
> agent1.sinks.hdfs-sink1.hdfs.filePrefix = event
> agent1.sinks.hdfs-sink1.hdfs.writeFormat = Text
> agent1.sinks.hdfs-sink1.hdfs.rollInterval = 60
> agent1.sinks.hdfs-sink1.hdfs.rollCount = 0
> agent1.sinks.hdfs-sink1.hdfs.rollSize = 0
> agent1.sinks.hdfs-sink1.hdfs.fileType = DataStream
> agent1.sinks.hdfs-sink1.hdfs.batchSize = 1000
> # Finally, now that we've defined all of our components, tell
> # agent1 which ones we want to activate.
> agent1.channels = ch1
> agent1.sources = exec-source1
> agent1.sinks = hdfs-sink1
> So far I only get about 20Mb/min or less than 1 Mb/sec. I am wondering how
> far it can be improved. Is there any Benchmark on HDFS Sink performance.
> Thanks in Advance,
Apache MRUnit - Unit testing MapReduce - http://incubator.apache.org/mrunit/
Pankaj Gupta 2012-11-09, 18:28
Pankaj Gupta 2012-11-09, 19:27