How fast is BonsaiDb?

WARNING: All of these benchmarks are currently wrong. Due to breaking changes on main, we have not updated the graphs yet. This page will be revised soon.

BonsaiDb's goal for performance is to be comparable to other popular, general-purpose, ACID-compliant databases such as SQLite, PostgreSQL, and MongoDB. We do not consider performance to be a core focus in development and will make decisions that create a better developer experience over raw performance.

What do these benchmarks tell us about BonsaiDb?

Benchmarking a database is challenging, as each database has a unique set of features. Microbenchmarks aren't good for evaluating performance, because often the performance of various functionality changes based on what other load the database is under. For example, read performance will be impacted by the amount of writes happening and vice-versa.

The Commerce Benchmark aims to provide a way to simulate different ratios of read operations and write operations for a simple eCommerce schema. It might be tempting to use the results of this benchmark to claim that BonsaiDb will be faster for your application than another database choice.

The unfortunate reality is that no matter how well designed a benchmark suite is, it will never be able to predict the performance of a database solution for your application with your usage patterns.

The only conclusion we want readers to draw from these benchmarks is: BonsaiDb's performance is comparable to other mainstream databases.

General Notes

All benchmarks on this page are tailored to try to give a fair comparison between what functionality BonsaiDb offers and the closest equivalent mode in the database being compared against. This means that there are ways to operate other databases in configurations that provide different guarantees (such as not being fully ACID-compliant) and sometimes those modes can yield faster execution. These benchmarks will always pick the best way to compare other databases against BonsaiDb's guarantees, and as such, will not give a complete picture of any of the databases being compared.

It is important to consider the hardware which BonsaiDb will be deployed to. Many developers have more powerful development machines than their applications eventually are deployed on. Each database has slightly different performance behaviors on various types of hard drives and memory constraints.

Commerce Benchmark

The Commerce Benchmark attempts to simulate a relational database workload. All benchmarks are measuring fully ACID-compliant reads and writes. Click on either graph for the full report.

This suite runs a series of operations with various configurations for:

For more information about this suite, including instructions on how to run the suite on your own hardware, see the benchmark's README.

Current Results

Results from a Scaleway GP1-XS instance running Ubuntu 20.04 with 4 CPU cores, 16GB of RAM, and local NVME storage (only occasionally updated):

Commerce Benchmark Overview from Scaleway

Results when executed on GitHub Actions:

Commerce Benchmark Overview from GitHub Actions

Microbenchmarks

BonsaiDb doesn't have many microbenchmarks yet. This suite uses Criterion for its measurements.

This suite currently only has one benchmark for documents/collections, and a few for the key-value store.

Current Results

Clicking on any graph will take you to the individual benchmark's report page.

ACID-compliant Document Inserts

These benchmarks measure the ACID-compliant insert speed of documents of varying sizes. Random data is utilized for the document contents, which is a worst-case scenario when enabling compression.

Results from a Scaleway GP1-XS instance:

save_documents overview from Scaleway

Results when executed on GitHub Actions:

save_documents overview from GitHub Actions

Key-Value Get Bytes

This benchmark measures the atomic get speed from the key-value store with keys of varying sizes.

Results from a Scaleway GP1-XS instance:

get-bytes overview from Scaleway

Results when executed on GitHub Actions:

get-bytes overview from GitHub Actions

Key-Value Set Bytes

This benchmark measures the atomic set (insert) speed from the key-value store with keys of varying sizes. These operations are not confirmed to be written to disk before the results are reported, making the key-value store not ACID compliant.

Results from a Scaleway GP1-XS instance:

set-bytes overview from Scaleway

Results when executed on GitHub Actions:

set-bytes overview from GitHub Actions

Key-Value Increment Value

This benchmark measures the atomic increment operation speed from the key-value store. These operations are not confirmed to be written to disk before the results are reported, making the key-value store not ACID compliant.

Results from a Scaleway GP1-XS instance:

get-bytes overview from Scaleway

Results when executed on GitHub Actions:

get-bytes overview from GitHub Actions