YCSB/geode

Quick Start

This section describes how to run YCSB on Apache Geode (incubating).

Get Apache Geode

You can download Geode from http://geode.incubator.apache.org/releases/

Start Geode Cluster

Use the Geode shell (gfsh) to start the cluster. You will need to start at-least one locator which is a member discovery service and one or more Geode servers.

Launch gfsh:

$ cd $GEODE_HOME
$ ./bin/gfsh

Start a locator and two servers:

gfsh> start locator --name=locator1
gfsh> start server --name=server1 --server-port=40404
gfsh> start server --name=server2 --server-port=40405
gfsh> configure pdx --read-serialized=true

Create the "usertable" region required by YCSB driver:

gfsh>create region --name=usertable --type=PARTITION

gfsh has tab autocompletion, so you can play around with various options.

Start YCSB workload

From your YCSB directory, you can run the ycsb workload as follows

./bin/ycsb load geode -P workloads/workloada -p geode.locator=host[port]

(default port of locator is 10334).

In the default mode, ycsb geode driver will connect as a client to the geode cluster. To make the ycsb driver a peer member of the distributed system use the property -p geode.topology=p2p -p geode.locator=host[port]

Note: For update workloads, please use the property -p writeallfields=true