About WekaIO System Performance Testing

There are 3 main performance metrics when measuring a storage system performance:

1. Latency, which is the time from operation initiation to completion.

2. The number of IO operations that the system can process concurrently.

3. The bandwidth of data that the system can process concurrently.

Each of these performance metrics apply to read operations, write operations or a mixture of read and write operations.

When measuring the WekaIO system performance, different mount modes produce different performance characteristics. Additionally, client network configuration (using either space networking or kernel UDP) also have a significant effect on performance.

The FIO Utility

The FIO Utility is a generic open source storage performance testing tool which can be defined as described here. In this documentation, the usage of FIO version 3.5 is assumed.

WekaIO System Performance Testing

Small Write Latency

Description

This test measures the minimal achievable write latency under a light load. The test measures the latency over a single threaded sequence of 4 KB writes across multiple files. Each write is executed only after the previous write has been acknowledged.

Preparation

export WEKA_MOUNT=/mnt/weka
export BENCHMARK_ID=FioWrites4KSingleThread
export WORKING_DIR=$WEKA_MOUNT/$BENCHMARK_ID
mkdir -p $WORKING_DIR

Run Benchmark

fio --name=$BENCHMARK_ID --directory=$WORKING_DIR --filesize=104857600 \\
     --group_reporting --numjobs=1 --ioengine=posixaio --nrfiles=32 --invalidate=0 \\
     --iodepth=1 --max-jobs=1 --rwmixread=0 --create_serialize=0 --runtime=70 \\
     --time_based --direct=1 --randrepeat=1 --filename_format='$jobnum/$filenum' \\
     --clat_percentiles=1 --blocksize=4096 --readwrite=randwrite

Example of Test Output

In this test output example, results show an average latency of 529 microseconds, where 99.5% of the writes terminated in 766 microseconds or less.

Small Read Latency

This test measures the minimal achievable read latency under a light load. The test measures the latency over a single threaded sequence of 4 KB reads across multiple files. Each read is executed only after the previous read has been served.

Preparation

export WEKA_MOUNT=/mnt/weka
export BENCHMARK_ID=FioReads4KSingleThread
export WORKING_DIR=$WEKA_MOUNT/$BENCHMARK_ID
mkdir -p $WORKING_DIR

Run Benchmark

fio --blocksize=4k --direct=1 --numjobs=1 --directory=$WORKING_DIR \\
    --ioengine=libaio --invalidate=0 --time_based --name=$BENCHMARK_ID --nrfiles=1 \\
    --clat_percentiles=1 --max-jobs=1 --randrepeat=1 --create_serialize=0 \\
    --group_reporting ' --filename_format=$jobnum/$filenum' --runtime=60 \\
    --rwmixread=0 --iodepth=256 --readwrite=randread --filesize=10485760000

Example of Test Output

In this test output example, results show an average latency of 224 microseconds, where 99.5% of the writes terminated in 338 microseconds or less.

Large Bandwidth Writes and Sequential Writes

Description

This test measures the overall system throughput for large (1 MB) writes, which is the same as large sequential writes. The scripts below will try to maximize the system write throughput from a single client. The test utilizes multiple threads, each one performing 1 MB writes.

Preparation

export WEKA_MOUNT=/mnt/weka
export BENCHMARK_ID=FioWrite1MMultiThread
export WORKING_DIR=$WEKA_MOUNT/$BENCHMARK_ID
mkdir -p $WORKING_DIR

Run Benchmark

fio --blocksize=1048576 --direct=1 --numjobs=32 --directory=$WORKING_DIR \\
    --ioengine=libaio --invalidate=0 --time_based --name=$BENCHMARK_ID --nrfiles=1 \\
    --clat_percentiles=1 --max-jobs=32 --randrepeat=1 --create_serialize=0 \\
    --group_reporting '--filename_format=$jobnum/$filenum' --runtime=70 --rwmixread=0 \\
    --iodepth=1 --readwrite=write --filesize=104857600

Example of Test Output

In this test output example, results show a bandwidth of 2.8 Gigabytes/second.

Large Bandwidth Reads and Sequential Reads

Description

This test measures the overall system throughput for large (1MB) reads, which is the same as sequential reads. The scripts below will try to maximize the system read throughput from a single client. The test utilizes multiple threads, each one performing 1MB reads.

Preparation

export WEKA_MOUNT=/mnt/weka
export BENCHMARK_ID=FioReads1MMultiThread
export WORKING_DIR=$WEKA_MOUNT/$BENCHMARK_ID
mkdir -p $WORKING_DIR

Run Benchmark

fio —blocksize=1048576 --direct=1 --numjobs=32 --directory=$WORKING_DIR \\
    --ioengine=libaio --invalidate=0 --time_based --name=$BENCHMARK_ID --nrfiles=1 \\
    --clat_percentiles=1 --max-jobs=32 --randrepeat=1 --create_serialize=0 \\
    --group_reporting '--filename_format=$jobnum/$filenum' --runtime=70 \\
    --rwmixread=0 --iodepth=1 --readwrite=read --file

Example of Test Output

In this test output example, results show a bandwidth of 2.8 Gigabytes/second.

IOPS Write

Description

This test measures the overall system ability to execute small (4 KB) writes. The scripts below will try to maximize the number of concurrent operations from a single client. The test utilizes multiple threads, each one performing 4 KB writes.

Preparation

export WEKA_MOUNT=/mnt/weka
export BENCHMARK_ID=IOPSWrite1MMultiThread
export WORKING_DIR=$WEKA_MOUNT/$BENCHMARK_ID
mkdir -p $WORKING_DIR

Run Benchmark

fio --blocksize=4k --direct=1 --numjobs=1 --directory=$WORKING_DIR \\
    --ioengine=libaio --invalidate=0 --time_based --name=$BENCHMARK_ID --nrfiles=1 \\
    --clat_percentiles=1 --max-jobs=1 --randrepeat=1 --create_serialize=0 \\
    --group_reporting '--filename_format=$jobnum/$filenum' --runtime=60 \\
    --rwmixread=0 --iodepth=256 --readwrite=randwrite --filesize=10485760000

Example of Test Output

In this test output example, results show an average IOPS of 127,402.

IOPS Read

Description

This test measures the overall system throughput for small (4 KB) reads. The scripts below will try to maximize the system read throughput from a single client. The test utilizes multiple threads, each one performing 4 KB reads.

Preparation

export WEKA_MOUNT=/mnt/weka
export BENCHMARK_ID=IOPSRead1MMultiThread
export WORKING_DIR=$WEKA_MOUNT/$BENCHMARK_ID
mkdir -p $WORKING_DIR

Run Benchmark

fio --blocksize=4k --direct=1 --numjobs=1 --directory=$WORKING_DIR \\
    --ioengine=libaio --invalidate=0 --time_based --name=$BENCHMARK_ID \\
    --nrfiles=1 --clat_percentiles=1 --max-jobs=1 --randrepeat=1 \\
    --create_serialize=0 --group_reporting '--filename_format=$jobnum/$filenum' \\
    --runtime=60 --rwmixread=0 --iodepth=256 --readwrite=randread \\
    --filesize=10485760000

Example of Test Output

In this test output example, results show an average IOPS of 127,402.

​