Thanks for the reply and sharing your experience, I will try the same.
Regards,
Bhavi Dixit
On Monday, 2 December 2013 08:13:16 UTC+5:30, Ross Simpson wrote:
I just completed some ES performance testing. Here's what I did:
Get some test queries. I looked at some production traffic, extracted
some common query types, and inserted pseudo-random data in place of
variables (e.g. user search terms).Instrumented the various components of my infrastructure (NewRelic and
SPM http://sematext.com/spm/index.html for Elasticsearch, NewRelic for
my application and the load generators).Used Siege to hammer the cluster with requests. I experimented with
various parameters, including concurrency, ramp-up, etc., in order to get
the right load profile.Evaluate the results from the instrumentation. Identify the components
which are the bottleneck.Partway through the test, I switched from Siege to JMeter, which has much
better reporting.The most important part of this was probably getting a good set of
representative data. I didn't have actual production ES queries to replay,
so I had to convert other queries to the ES format.My cluster of ES nodes got to about 1700 QPS, and ES never broke a sweat
-- the bottleneck was in my application. I'm confident that ES could
handle 2-3 times more traffic than that before it started to work hard.Ross
On Sunday, 1 December 2013 19:16:15 UTC+11, Bharvi Dixit wrote:
Hi,
Could anyone tell me how can i do a load testing on my elasticsearch
cluster.?
I want to test how many concurrent user requests it can handle.Regards
Bharvi Dixit
--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/aa1660c1-d321-44ac-a8af-c58eb8385593%40googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.