# [Alternative performance testing of MongoDB](https://learn.arm.com/learning-paths/servers-and-cloud-computing/mongodb/perf_mongodb/)

## In this learning path

- [Introduction](https://learn.arm.com/learning-paths/servers-and-cloud-computing/mongodb/)
- [Install MongoDB on Arm](https://learn.arm.com/learning-paths/servers-and-cloud-computing/mongodb/run_mongodb/)
- [Creating MongoDB test scenarios](https://learn.arm.com/learning-paths/servers-and-cloud-computing/mongodb/create_replica_set/)
- [Benchmark MongoDB on Arm with Yahoo Cloud Serving Benchmark (YCSB)](https://learn.arm.com/learning-paths/servers-and-cloud-computing/mongodb/benchmark_mongodb-8.0/)
- [Three node replica set testing with YCSB](https://learn.arm.com/learning-paths/servers-and-cloud-computing/mongodb/replica_set_testing/)
- [Alternative performance testing of MongoDB](https://learn.arm.com/learning-paths/servers-and-cloud-computing/mongodb/perf_mongodb/)
- [Automate MongoDB Performance Benchmarking Infrastructure Setup with Pulumi](https://learn.arm.com/learning-paths/servers-and-cloud-computing/mongodb/automate_setup_pulumi/)
- [Next Steps](https://learn.arm.com/learning-paths/servers-and-cloud-computing/mongodb/_next-steps/)

To measure the performance of MongoDB, use the [MongoDB performance test tool](https://github.com/idealo/mongodb-performance-test).  
This is an open source Java application that tests the MongoDB performance, such as latency and throughput, by running one or more threads executing either all the same or different database operations, such as Inserts, Updates, Deletes, Counts or Finds until a defined number of operations is executed or a defined maximum runtime is reached.

## Install OpenJDK packages

Install the appropriate run-time environment to be able to use the performance test tool.

```bash
sudo apt install default-jre default-jdk -y
```

```bash
sudo yum install java-17-openjdk
```

For more information see the [OpenJDK](https://openjdk.org/install/) website.

## Setup the MongoDB performance test tool

On your instance running MongoDB (you may need to start a new terminal), clone the `MongoDB performance test tool` project:

```bash
git clone https://github.com/idealo/mongodb-performance-test.git
```

Now change into the project folder and execute the JAR file to see its usage instructions:

```bash
cd mongodb-performance-test
java -jar ./latest-version/mongodb-performance-test.jar
```

This will print a description of how to use the Java application.

## Run Insert test

Run a test that inserts documents on `localhost:27017` (default).

First, set an environment variable for the JAR file path for convenience:

```bash
export jarfile=./latest-version/mongodb-performance-test.jar
```

Use the following options:

- `-m` defines the test mode (e.g., `insert`, `update_one`).
- `-o` defines the number of operations (iterations).
  - Alternatively, use `-d` to specify a duration limit (in seconds).
- `-t` defines the number of threads.
- `-db` defines the database to use.
- `-c` defines the collection to use.

For example, run an insert test for 1 million operations using 10 threads:

```bash
java -jar $jarfile -m insert -o 1000000 -t 10 -db test -c perf
```

As the test runs, the progress count will be printed periodically. It will increase until it reaches 1 million, and then the test will end.

## Run Update-one test

Similarly, to run an update test (updating one document per query) using 10, 20, and finally 30 threads for 1 hour each (3 hours total), run the following command:

```bash
java -jar $jarfile -m update_one -d 3600 -t 10 20 30 -db test -c perf
```

For instructions on running any other tests or more details on the metrics reported, refer to the [github project for the performance tool](https://github.com/idealo/mongodb-performance-test#readme).

## View the results

During each test, statistics over the last second are printed to the console every second. After the test completes, final summary statistics are displayed. The following is example output from the end of the Insert test run:

```text
__output__ -- Timers ----------------------------------------------------------------------
__output__ stats-per-run-INSERT
__output__ 2022-07-05 19:14:45,894 [main] INFO  d.i.mongodb.perf.MongoDbAccessor - <<< closeConnections localhost:27017
__output__           count = 1000000
__output__       mean rate = 5001.61 calls/second
__output__   1-minute rate = 5042.28 calls/second
__output__   5-minute rate = 3699.92 calls/second
__output__  15-minute rate = 2963.07 calls/second
__output__         min = 0.36 milliseconds
__output__         max = 15.59 milliseconds
__output__        mean = 1.93 milliseconds
__output__      stddev = 0.66 milliseconds
__output__      median = 1.87 milliseconds
__output__        75% <= 1.99 milliseconds
__output__        95% <= 2.22 milliseconds
__output__        98% <= 2.64 milliseconds
__output__        99% <= 3.85 milliseconds
__output__      99.9% <= 15.59 milliseconds
```

Detailed per-second metrics are also output to a CSV file named `stats-per-second-[mode].csv` (e.g., `stats-per-second-INSERT.csv`), located in the same folder as the JAR file. `[mode]` corresponds to the executed mode(s), such as `INSERT`, `UPDATE_ONE`, `DELETE_ONE`, etc.