Standard query benchmark?

Elastic Stack Elasticsearch
1

Hi,

Are there any standard-ish search benchmarking mechanisms in ES?
I'm after running something that would allow me to compare "baseline
queries" vs. the same queries with bloom filter turned on vs. the same
queries with pulsing codec.
Unless there is something standard-ish I could use, I'm tempted to just do
this:

If there are better ways to compare baseline vs. bloom filter vs, pulsing
codec, I'm all eyeballs!

Thanks,
Otis

ELASTICSEARCH Performance Monitoring - http://sematext.com/spm/index.html

--
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.
For more options, visit https://groups.google.com/groups/opt_out.

2

there are no benchmark tools yet. I think you are stuck with what you lined
up below.
I am not sure if benchmarking is really needed at this point since unless
you do a lot of key / value lookups those codecs won't buy you much and
default codec pulses by default for very low docfreq terms.

simon

On Friday, March 22, 2013 6:22:18 AM UTC+1, Otis Gospodnetic wrote:

Hi,

Are there any standard-ish search benchmarking mechanisms in ES?
I'm after running something that would allow me to compare "baseline
queries" vs. the same queries with bloom filter turned on vs. the same
queries with pulsing codec.
Unless there is something standard-ish I could use, I'm tempted to just do
this:

If there are better ways to compare baseline vs. bloom filter vs, pulsing
codec, I'm all eyeballs!

Thanks,
Otis

ELASTICSEARCH Performance Monitoring - Sematext Monitoring | Infrastructure Monitoring Service

--
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.
For more options, visit https://groups.google.com/groups/opt_out.

3

Hi Simon,

Aha, thanks for the info! If default already pulses just like what an
explicit pulsing codec would do then maybe we can ignore that..... but
still want to test bloom filter vs. the default. Indeed, lots of key
lookups, that's what the queries we're looking at now are like.

Otis

ELASTICSEARCH Performance Monitoring - Sematext Monitoring | Infrastructure Monitoring Service

On Friday, March 22, 2013 2:39:37 PM UTC-4, simonw wrote:

there are no benchmark tools yet. I think you are stuck with what you
lined up below.
I am not sure if benchmarking is really needed at this point since unless
you do a lot of key / value lookups those codecs won't buy you much and
default codec pulses by default for very low docfreq terms.

simon

On Friday, March 22, 2013 6:22:18 AM UTC+1, Otis Gospodnetic wrote:

Hi,

Are there any standard-ish search benchmarking mechanisms in ES?
I'm after running something that would allow me to compare "baseline
queries" vs. the same queries with bloom filter turned on vs. the same
queries with pulsing codec.
Unless there is something standard-ish I could use, I'm tempted to just
do this:

If there are better ways to compare baseline vs. bloom filter vs, pulsing
codec, I'm all eyeballs!

Thanks,
Otis

ELASTICSEARCH Performance Monitoring - Sematext Monitoring | Infrastructure Monitoring Service

--
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.
For more options, visit https://groups.google.com/groups/opt_out.

4

On Saturday, March 23, 2013 4:43:30 AM UTC+1, Otis Gospodnetic wrote:

Hi Simon,

Aha, thanks for the info! If default already pulses just like what an
explicit pulsing codec would do then maybe we can ignore that..... but
still want to test bloom filter vs. the default. Indeed, lots of key
lookups, that's what the queries we're looking at now are like.

FYI - Elasticsearch already uses bloom filter for the uid field. Using a
bloom codec is only worth it when the likelyhood of a miss (the term is
likely not in the term dict or only in one segment, just like uids) is
very high. You are saving like a couple of micro-seconds and only if you
really do this a lot like applying deletes in the IndexWriter this is worth
the trouble. I totally wanna encourage you to try this and share your
results, just wanna provide you with the info ahead of time.

simon

Otis

ELASTICSEARCH Performance Monitoring - Sematext Monitoring | Infrastructure Monitoring Service

On Friday, March 22, 2013 2:39:37 PM UTC-4, simonw wrote:

there are no benchmark tools yet. I think you are stuck with what you
lined up below.
I am not sure if benchmarking is really needed at this point since unless
you do a lot of key / value lookups those codecs won't buy you much and
default codec pulses by default for very low docfreq terms.

simon

On Friday, March 22, 2013 6:22:18 AM UTC+1, Otis Gospodnetic wrote:

Hi,

Are there any standard-ish search benchmarking mechanisms in ES?
I'm after running something that would allow me to compare "baseline
queries" vs. the same queries with bloom filter turned on vs. the same
queries with pulsing codec.
Unless there is something standard-ish I could use, I'm tempted to just
do this:

If there are better ways to compare baseline vs. bloom filter vs,
pulsing codec, I'm all eyeballs!

Thanks,
Otis

ELASTICSEARCH Performance Monitoring -
Sematext Monitoring | Infrastructure Monitoring Service

--
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.
For more options, visit https://groups.google.com/groups/opt_out.