The compression review tool provides a compressibility metric (the maximum compression possible on each collection's documents) by sampling the actual documents and applying the requested compression algorithm. The actual compression achieved will be a lower number due to fragmentation and write amplification.
The tool samples 1000 documents in each collection to determine the compressibility of the data. A larger number of documents can be sampled via the --sample-size parameter.
- Python 3.7+
- pymongo Python package - tested versions
- MongoDB 2.6 - 3.4 | pymongo 3.10 - 3.12
- MongoDB 3.6 - 5.0 | pymongo 3.12 - 4.0
- MongoDB 5.1+ | pymongo 4.0+
- DocumentDB | pymongo 3.10+
- If not installed - "$ pip3 install pymongo"
- lz4 Python package
- If not installed - "$ pip3 install lz4"
- zstandard Python package
- If not installed - "$ pip3 install zstandard"
python3 compression-review.py --uri <server-uri> --server-alias <server-alias>
- Default compressions tested is lz4/fast/level1 and zstandard/level3/4K/Dictionary
- To test other compression techniques provide --compressor <compression-type>
- Run on any instance in the replica set
- Use a different <server-alias> for each server analyzed, output file is named using <server-alias> as the starting portion
- Creates a single CSV file per execution (so default creates two)
- The <server-uri> options can be found at https://www.mongodb.com/docs/manual/reference/connection-string/
- If your URI contains ampersand (&) characters they must be escaped with the backslash or enclosed your URI in double quotes
- For DocumentDB use either the cluster endpoint or any of the instance endpoints
This tool is licensed under the Apache 2.0 License.