Benchmarking accessing ElasticSearch via JRuby

bench_direct.rb

#!/usr/bin/env ruby
require 'java'

import org.elasticsearch.client.transport.TransportClient
import org.elasticsearch.common.transport.InetSocketTransportAddress
import org.elasticsearch.common.settings.ImmutableSettings
import org.elasticsearch.action.search.SearchResponse
import org.elasticsearch.action.search.SearchType
import org.elasticsearch.index.query.FilterBuilders
import org.elasticsearch.index.query.QueryBuilders

def search(client)
  response = client.prepareSearch("twitter")
                   .setTypes("tweet")
                   .setQuery(QueryBuilders.termQuery("trying", "out"))               # Query
                   .setFrom(0).setSize(10)
                   .execute()
                   .actionGet()
  p response.getHits().totalHits()
end

def index(client)
  json = "{" +
         "\"user\":\"kimchy\"," +
         "\"postDate\":\"2013-01-30\"," +
         "\"message\":\"trying out Elasticsearch #{Time.now.to_f}\"" +
         "}"

  client.prepareIndex("twitter", "tweet")
        .setSource(json)
        .execute()
        .actionGet()
end

settings = ImmutableSettings.settingsBuilder()
                            .put("cluster.name", "elasticsearch_test").build()

client = TransportClient.new(settings)
client.addTransportAddress(InetSocketTransportAddress.new('localhost', 9300))

(ENV['SEARCH_TIMES'].to_i).times do
  search(client)
end

(ENV['INDEX_TIMES'].to_i).times do
  index(client)
end

client.close

log4j.properties

# Set root logger level to DEBUG and its only appender to A1.
log4j.rootLogger=DEBUG, A1

# A1 is set to be a ConsoleAppender.
log4j.appender.A1=org.apache.log4j.ConsoleAppender

# A1 uses PatternLayout.
log4j.appender.A1.layout=org.apache.log4j.PatternLayout
log4j.appender.A1.layout.ConversionPattern=%-4r [%t] %-5p %c %x - %m%n

results.md

export CLASSPATH=$(ls ~/Downloads/elasticsearch-1.0.0/lib/*.jar | tr '\n' ':')

On a rMBP (8GB RAM, 2.3 GHz Intel Core i7, 256GB SSD)

Jruby
Index	Warm up	1,000	docs/sec	10,000	docs/sec	100,000	docs/sec	1,000,000	docs/sec
	2.35	3.13	319.49	6.55	1526.72	25.99	3847.63	250.52	3991.70
	2.32	3.08	324.68	6.22	1607.72	25.20	3968.25		#DIV/0!
Search (100K)		3.51				25.63	3901.68
		3.28				25.28	3955.70
MRI 2.1.0 + rest-client
Index	Warm up	1,000	docs/sec	10,000	docs/sec	100,000	docs/sec	1,000,000	docs/sec
	0.3	1.72		12.66
	0.31	1.49		9.29
		1.37		10.14

Initial results show JRuby ~2x faster for larger workloads