gistfile2.js

> db.ticks.findOne({'_id.cid': 732814})
{
	"_id" : {
		"cid" : 732814,
		"ts" : 1245442500,
		"drn" : 60
	},
	"vals" : [
		[
			3622,
			0,
			4
		]
	]
}


# The "vals" array may have several thousand 3-element arrays inside of it

gistfile3.txt

# Running the ruby program


Mac-Pro:perf cremes$ rvm use default
Using /Users/cremes/.rvm/gems/ruby-1.9.2-p0
Mac-Pro:perf cremes$ time ruby test_mongo.rb
total ticks [155816736]
 46.500000   9.390000  55.890000 (166.054945)

real	2m56.354s
user	0m46.640s
sys	0m9.461s

# Ran with an average of 100% cpu

gistfile4.txt

# Running the java program using the latest master (2.1)


Mac-Pro:examples cremestime $ java -classpath ../mongo.jar:../bson.jar:. Perf 
total ticks [155816736]

real	9m43.431s
user	13m58.968s
sys	0m48.477s


# Ran with an average of 160% cpu

gistfile5.txt

# Running the Ruby program with JRuby


Mac-Pro:perf cremes$ rvm use jruby-1.5.1
Using /Users/cremes/.rvm/gems/jruby-1.5.1
Mac-Pro:perf cremes$ time ruby test_mongo.rb
total ticks [155816736]
544.005000   0.000000 544.005000 (544.005000)

real	9m16.736s
user	16m23.281s
sys	0m47.347s

Perf.java

import com.mongodb.Mongo;
import com.mongodb.DBCollection;
import com.mongodb.BasicDBObject;
import com.mongodb.BasicDBList;
import com.mongodb.DBObject;
import com.mongodb.DBCursor;
import com.mongodb.DB;

import java.util.Set;
import java.util.List;

public class Perf {

    public static void main(String[] args) throws Exception {

        // connect to the local database server
        Mongo m = new Mongo( "192.168.0.104" );

        // get handle to "mydb"
        DB db = m.getDB( "market_data" );
    
        // get a collection object to work with
        DBCollection coll = db.getCollection("ticks");
        
	// query for a specific contract ID
	BasicDBObject contract_query = new BasicDBObject();
	contract_query.put("_id.cid", 732814);

        // request all ticks for a specific contract ordered by timestamp
        DBCursor cur = coll.find(contract_query).sort( new BasicDBObject( "_id.ts", 1 ));
	BasicDBList doc;
	int	count = 0;

        while(cur.hasNext()) {
	    doc = (BasicDBList) cur.next().get("vals");
            count += doc.size();
        }

        System.out.println("total ticks [" + count + "]");
    }
}

perf.rb

require 'rubygems'
require 'mongo'
require 'benchmark'


connection = Mongo::Connection.new '192.168.0.104'
db = connection.db "market_data"
collection = db.collection 'ticks'

cid = 732814

cursor = Mongo::Cursor.new(collection,
{
  :timeout => false,
  :limit => 0,
  :skip => 0,
  :order => [['_id.ts', Mongo::ASCENDING]],
  :selector => {
    '_id.cid' => cid
  }
}
)

total_ticks = 0

result = Benchmark.measure do
  while cursor.has_next?
    doc = cursor.next_document
    total_ticks += doc['vals'].size
  end
end

puts "total ticks [#{total_ticks}]"
puts result

Rerunning the Java program with a larger heap and -server gave some better results.

Mac-Pro:examples cremes$ time java -Xmx2g -server -classpath ../mongo.jar:../bson.jar:. Perf
total ticks [155816736]

real 7m5.905s
user 5m40.339s
sys 0m9.588s

This is better but still far slower than the Ruby C extension.

Also, reran with 2.1 release with -server and more heap and did NOT notice any performance difference. It ran the same as 2.1-master (or 2.2, whatever).

Mac-Pro:examples cremes$ java -version
java version "1.6.0_20"
Java(TM) SE Runtime Environment (build 1.6.0_20-b02-279-10M3065)
Java HotSpot(TM) 64-Bit Server VM (build 16.3-b01-279, mixed mode)