Benchmarking SurrealDB 3.x vs. Postgres, Mongo, Neo4j and Redis (With Fsync)

Benchmark tests conducted on the same hardware and configurations show that SurrealDB 3.x outperforms its previous versions and rivals leading databases like Postgres and MongoDB in key performance metrics, with full durability enabled. This development highlights SurrealDB’s growing maturity as a multi-model database suitable for production environments.

The benchmarks used an AMD Ryzen Threadripper 9970X with 128 GiB RAM, NVMe storage, and Ubuntu 24.04. All databases were configured with production-grade settings, including fsync enabled, to reflect real-world durability requirements. The workloads involved 128 clients issuing 48 concurrent queries on datasets of 5-15 million rows, covering CRUD operations, full-table scans, and indexed lookups.

SurrealDB 3.x showed a 136% increase in indexed query speed and an 11,894% improvement in full-table scan performance compared to earlier versions. Its CRUD throughput reached 141,000 operations per second, a 31% boost over SurrealDB 2.x. The query planner and storage engine overhaul eliminated per-row decoding overhead, reducing query times from minutes to seconds.

Compared to PostgreSQL, SurrealDB was approximately 1.5 times faster on create, update, and delete operations, although Postgres maintained an edge in raw read performance. Against MongoDB, SurrealDB was roughly 1.3 times faster on reads and 5-7 times faster on write operations. The benchmarks also included comparisons with Neo4j and Redis, emphasizing SurrealDB’s versatility across different data models.

Why It Matters

This benchmarking effort demonstrates SurrealDB’s rapid performance improvements and its competitiveness across relational, document, and graph workloads. The full durability configuration aligns with production standards, making SurrealDB a more viable option for applications requiring reliable, high-throughput multi-model data handling. The results could influence database selection decisions for developers seeking an integrated solution that balances performance and durability.

Background

Previous benchmark rounds with fsync disabled showed faster performance but did not reflect typical production environments. SurrealDB’s internal overhaul between versions 2.x and 3.x has yielded substantial performance gains. The comparison on the same hardware and configurations provides an apples-to-apples view, emphasizing the importance of durability settings for realistic assessments. The benchmarks are part of ongoing efforts to position SurrealDB as a serious alternative to established databases in multi-model use cases.

“The performance improvements in SurrealDB 3.x are a direct result of fundamental internal rebuilds, including query, parser, and storage layer enhancements.”

— SurrealDB team

“Using identical hardware and production-grade configurations, SurrealDB demonstrated significant speedups, especially in full-table scans and indexed queries.”

— Benchmark author

What Remains Unclear

It remains unclear how SurrealDB will perform under different hardware setups, with varying workload mixes, or in real-world production environments beyond controlled benchmarks. Long-term stability and scalability at larger scales are also still to be tested.

What’s Next

Further testing across diverse workloads, larger datasets, and real-world deployment scenarios is expected. SurrealDB’s development team plans to optimize query planner and storage engine features further and release version 3.1, which aims to close the remaining performance gaps with traditional relational databases.

Key Questions

How does SurrealDB 3.x compare to other databases in real-world use?

While benchmarks show promising results, real-world performance depends on workload specifics, hardware, and configuration. Further testing is needed to confirm these gains in production environments.

What are the main improvements in SurrealDB 3.x?

The major updates include a rebuilt query planner, optimized storage engine, and elimination of per-row decoding overhead, leading to faster query execution and full-table scans.

Is full durability enabled in these benchmarks typical for production?

Yes, the tests used fsync enabled, reflecting standard durability requirements in production systems, which often slow down performance but ensure data integrity.

When will SurrealDB 3.1 be available?

The SurrealDB team has announced plans to release version 3.1 soon, which will focus on further performance optimization and feature enhancements.

Source: Hacker News