In the benchmark tests, the producer (the client in the client/server example) runs a number of sampling iterations, each composed of a number of put() operations, with each put() operation writing a specific number of bytes. The settings used are reported to standard out at the start of each run. The default run uses 60 sampling iterations, with 5000 put operations per sample (thus the total 300,000 puts reported by the example), and each operation puts 10 kilobytes of data.
Run time is roughly determined by the number of samples multiplied by the number of operations per sample. The higher these combined properties are set, the longer the run time.
You can modify the benchmark to approximate the system you plan to implement. You can pass any combination of these properties to the producer:
|benchmark.number-of-samples||Number of sampling iterations run by the producer. Default: 60|
|benchmark.operations-per-sample||Number of put operations performed for each sampling iteration. Default: 5000|
|benchmark.payload-bytes||Number of bytes of data cached by each put() operation. Default: 10240 (10 kilobytes).|
For example, this client invocation sets the payload to 100 kilobytes, leaving the defaults for the other properties:
$ java -cp "$GEMFIRE/SampleCode/quickstart/classes:$GEMFIRE/lib/server-dependencies.jar" -Dbenchmark.payload-bytes=102400 quickstart.BenchmarkClient