Document Mini Preview

Benchmark

GridGain/Apache Ignite 1.5.0 vs Hazelcast 3.6 Benchmark

Introduction

This is a comparison between GridGain 7.4/Apache Ignite 1.5.0 and Hazelcast 3.6 prepared using GridGain’s own benchmarking tool, Yardstick. As with our Hazelcast 3.6-Snapshot GridGain-7.4 Benchmark – EC2 benchmark, Hazelcast is faster in almost all benchmarks with less latency.

Test Environment

Specification Description
Cloud AWS EC2
EC2 Instance Type c4.2xlarge
vCPU 8
RAM 15GiB
AWS AMI http://thecloudmarket.com/image/ami-b66ed3de–amzn-ami-hvm-2014-09-1-x86-64-ebs
Operating System Linux
Java Oracle 1.7.0_55

Test Parameters

Parameter Value
Number of Nodes 4
Yardstick Setup 3 server member nodes 1 driver member node (4 node cluster total)
Number of Threads 64
Test Duration (minutes) 5
Benchmark Framework* Build yardstick from master https://github.com/hazelcast/yardstick, then the 2 dependent repos https://github.com/hazelcast/yardstick-ignite https://github.com/hazelcast/yardstick-hazelcast

Differences to GridGain/Ignite Yardstick

We are synced with GridGain and Ignite Yardstick repos as they stood at 2 March 2015, except for 3 changes which we have submitted as pull requests: https://github.com/gridgain/yardstick/pull/10

The specific commit hashes matching our forks of their repos are:

These allow a variable size payload to be used. GridGain/ignite are testing only with 1 byte which is lower than anyone would actually use for a value.

BenchMark Results, Logs, config

https://s3.amazonaws.com/dannyc/yardstick/2016_03_02-13_26_06-hz-3.6-vs-ig-1.5.0-final-driverMember.zip

The Results

Ignite cache put vs hazelcast map set (both sync backups, value size of 1 byte)

Benchmark 1: JCache Put

Ignite cache put/get vs hazelcast map set/get (both sync backups, value size of 1 byte)

Benchmark 2: JCache Put/Get

Ignite cache put vs hazelcast map set (both sync backups, value size of 1000 byte)

Benchmark 3: Put Tx

Ignite cache put/get vs hazelcast map set/get (both sync backups, value size of 1000 byte)

Benchmark 4: Put Get Tx Pessimistic

Ignite cache put vs hazelcast cache put (both sync backups, value size of 1 byte)

Benchmark 5: Put Get Tx Optimistic

Ignite cache put/get vs hazelcast cache put/get (both sync backups, value size of 1 byte)

Benchmark 6: SQL Query

Ignite cache put vs hazelcast map set (both Async backups, value size of 1 byte)

Benchmark 7: SQL Query & Put

Ignite cache Sql Query vs hazelcast map Sql Query (both sync backups)

Benchmark 8: Map Atomic Put

Ignite cache Sql Query with simultaneous put’s vs hazelcast map Sql Query with simultaneous set’s (both sync backups)

Benchmark 9: Map Atomic Put Get

Ignite cache implicit transactional put vs hazelcast map transactional sets (both sync backups, value size 1 byte)

Benchmark 9: Map Atomic Put Get

Ignite cache optimistic transactional put/get vs hazelcast map optimistic transactional sets/get (both sync backups, value size 1 byte)

Benchmark 9: Map Atomic Put Get

Ignite cache pessimistic transactional put/get vs hazelcast map pessimistic transactional sets/get (both sync backups, value size 1 byte)

Benchmark 9: Map Atomic Put Get

Get the Benchmark

Oops!

There's supposed to be a form right here, but its been hidden by your adblocker. Please disable your adblocker so you can get the benchmark you came for.

Hazelcast.com

Menu