# redis

> Database benchmark workloads: Redis latency, Redis server+client speed, Redis server (extrapolated) speed.

Database

## Redis latency {#redis-latency}

Running a pair of redis server and benchmarking client (memtier\_benchmark) on each vCPU to evaluate the performance of SET operations, using different number of concurrent pipelined requests. The average latency reported by memtier\_benchmark.

Benchmark details<dl><dt>Benchmark ID</dt><dd><code>redis:latency</code></dd><dt>Unit of measurement</dt><dd>Milliseconds (ms)</dd><dt>Server types evaluated</dt><dd>3,720</dd><dt>Total measurements</dt><dd>22,320</dd><dt>Range of measurements</dt><dd>0.25 – 750.7 <small>ms</small></dd><dt>Better</dt><dd>↓ Lower is better</dd></dl>

Configuration options

| Field | Description | Example values |
| --- | --- | --- |
| `pipeline` | The number of concurrent pipelined requests. | `1` `4` `16` `64` `256` `512` |
| `operation` | Type of operation, e.g. SET or GET a key. | `SET` |

:::warning

Latency is recorded during the redis:rps run, not as a standalone benchmark. As it reflects a characteristic of that throughput measurement, it is not recommended for comparing instances on its own; use it instead as a filter when interpreting the related throughput scores.

:::

Score distribution <small>(22,320 measurements)</small>

![Score distribution of Redis latency](/benchmarks/histograms/redis-latency.svg)

## Redis server+client speed {#redis-rps}

Running a pair of redis server and benchmarking client (memtier\_benchmark) on each vCPU to evaluate the performance of SET operations, using different number of concurrent pipelined requests. The measured RPS (ops/sec) is the sum of RPS measured in all parallel processes, but is not the maximum expected redis server speed, as the server(s) share CPU with the client(s).

Benchmark details<dl><dt>Benchmark ID</dt><dd><code>redis:rps</code></dd><dt>Unit of measurement</dt><dd>Operations per second (ops/sec)</dd><dt>Server types evaluated</dt><dd>3,720</dd><dt>Total measurements</dt><dd>22,320</dd><dt>Range of measurements</dt><dd>5K – 254.8M <small>ops/sec</small></dd><dt>Better</dt><dd>↑ Higher is better</dd></dl>

Configuration options

| Field | Description | Example values |
| --- | --- | --- |
| `pipeline` | The number of concurrent pipelined requests. | `1` `4` `16` `64` `256` `512` |
| `operation` | Type of operation, e.g. SET or GET a key. | `SET` |

Score distribution <small>(22,320 measurements)</small>

![Score distribution of Redis server+client speed](/benchmarks/histograms/redis-rps.svg)

## Redis server (extrapolated) speed {#redis-rps-extrapolated}

Running a pair of redis server and benchmarking client (memtier\_benchmark) on each vCPU to evaluate the performance of SET operations, using different number of concurrent pipelined requests. The extrapolated server speed is based on the measured speed adjusted by the server's and client's time spent executing in user/system mode, so trying to control for the client resource usage.

Benchmark details<dl><dt>Benchmark ID</dt><dd><code>redis:rps-extrapolated</code></dd><dt>Unit of measurement</dt><dd>Operations per second (ops/sec)</dd><dt>Server types evaluated</dt><dd>3,720</dd><dt>Total measurements</dt><dd>22,320</dd><dt>Range of measurements</dt><dd>10.5K – 555.7M <small>ops/sec</small></dd><dt>Better</dt><dd>↑ Higher is better</dd></dl>

Derived from [`redis:rps`](#redis-rps): Measured RPS adjusted by server/client user+system CPU time to control for client resource usage.

Configuration options

| Field | Description | Example values |
| --- | --- | --- |
| `pipeline` | The number of concurrent pipelined requests. | `1` `4` `16` `64` `256` `512` |
| `operation` | Type of operation, e.g. SET or GET a key. | `SET` |

Score distribution <small>(22,320 measurements)</small>

![Score distribution of Redis server (extrapolated) speed](/benchmarks/histograms/redis-rps-extrapolated.svg)
