Support Questions
Find answers, ask questions, and share your expertise
Announcements
Alert: Welcome to the Unified Cloudera Community. Former HCC members be sure to read and learn how to activate your account here.

What does nclients option of performance evaluation tool pe of hbase does?

Solved Go to solution
Highlighted

What does nclients option of performance evaluation tool pe of hbase does?

New Contributor

Hbase provides pe tool for performance benchmarking. It has an option to mention the number of clients. Is it the number of Hbase region servers we have in the cluster?

1 ACCEPTED SOLUTION

Accepted Solutions

Re: What does nclients option of performance evaluation tool pe of hbase does?

Hi @Aishwarya Dixit, pe or "Performance Evaluation" is a tool based on MapReduce to test read/writes to HBase. nclients means that 10*nclients mappers will be started to run the supplied pe command. Example:

hbase pe randomWrite 2
...
2017-09-02 01:31:17,681 INFO  [main] mapreduce.JobSubmitter: number of splits:20

starts a MR job with 20 mappers and 1 reducer. So you can start with a small number like 1-3 to make sure HBase works as expected, and then increase it to about max number of mappers you can run on your cluster divided by 10. Of course you can use a larger number, but then mappers will run in multiple "waves".

2 REPLIES 2

Re: What does nclients option of performance evaluation tool pe of hbase does?

New Contributor

@Jay SenSharma
Hi, Can you please help me in this regard?

Re: What does nclients option of performance evaluation tool pe of hbase does?

Hi @Aishwarya Dixit, pe or "Performance Evaluation" is a tool based on MapReduce to test read/writes to HBase. nclients means that 10*nclients mappers will be started to run the supplied pe command. Example:

hbase pe randomWrite 2
...
2017-09-02 01:31:17,681 INFO  [main] mapreduce.JobSubmitter: number of splits:20

starts a MR job with 20 mappers and 1 reducer. So you can start with a small number like 1-3 to make sure HBase works as expected, and then increase it to about max number of mappers you can run on your cluster divided by 10. Of course you can use a larger number, but then mappers will run in multiple "waves".