

Benchmarks

Nakama has been benchmarked under various workload types. Below you'll find parts of the benchmark results.

Get in touch if you'd like more exhaustive benchmark information.

Performance advantages stem from Nakama's modern hardware-friendly and ultra-scalable architecture. As a result, Nakama's performance grows as the hardware size grows.

Scaling both up and out offers many advantages: from simplified cluster management to access to generally better hardware and economies of scale.

Benchmark setup

These server workloads demonstrate single Nakama server cluster performance with a single server cluster for CockroachDB. The Tsung high-performance benchmark framework is used to run all workloads. All workloads are run with no warmups.

All workloads are conducted on Google Compute Engine.

Tsung server
OS	Ubuntu 16.04 LTS
Instance Type	n1-standard-4
CPU / Mem	4 / 15GB

Nakama server
OS	Ubuntu 16.04 LTS
Version	Nakama 1.0.1
Instance Type	n1-standard-1
CPU / Mem	1 / 3.75GB

CockroachDB server
OS	Ubuntu 16.04 LTS
Version	Nakama 1.0.4
Instance Type	n1-standard-2
CPU / Mem	2 / 7.5GB

Workload #1 - Register new user

This workload shows write operations via Nakama to the database server to register a new user.

Summary

A single Nakama server can handle mean loads of 400 requests/sec with requests served in 6.23 msec (mean) with a database write operation for a new user.

Registration benchmark results

Workload #2 - Login a user

This workload shows read operations via Nakama to the database server to login a user.

Summary

A single Nakama server can handle mean loads of 900 requests/sec with requests served in 37.67 msec (mean) with a database read operation for a user.

Login benchmark results

Workload #3 - CCU counts

This workload shows sustained concurrent connected clients which hold a socket open for 100 seconds each.

Summary

A single Nakama server can handle a peak CCU count of 11,000 users.

CCU benchmark results