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

Recent comprehensive benchmarks demonstrate that SurrealDB 3.x outperforms Postgres, MongoDB, Neo4j, and Redis in various workloads when configured for full durability, running on identical hardware. These results highlight significant performance gains, especially in CRUD operations and full-table scans, marking a notable advancement for SurrealDB.

The benchmarks were conducted on an AMD Ryzen Threadripper 9970X system with 128 GiB of DDR5 RAM and NVMe storage, using the same open-source benchmarking harness across all databases. Each database was tuned for production-grade durability, enabling fsync and WAL flushes, to reflect real-world deployment conditions. The workloads involved 128 clients executing 48 concurrent queries against datasets with 5 to 15 million rows, including mixed data types and complex queries.

SurrealDB 3.x showed a 164-fold increase in full-table scan performance compared to its previous version, with the latest version reaching 11 ops/sec. CRUD throughput improved by 31% from SurrealDB 2.x, with mean operations per second rising from 107k to 141k. Latency reductions ranged from 27% to 99%, depending on the workload. When compared to other databases, SurrealDB demonstrated faster write operations than Postgres and MySQL, with 1.5× and 5-7× higher throughput respectively. In document-oriented workloads, SurrealDB was approximately 1.3× faster than MongoDB in reads and significantly outperformed in unindexed filter scans, being roughly 2.7× faster.

Why It Matters

This benchmarking effort provides a realistic, production-oriented comparison of SurrealDB 3.x against established databases, emphasizing its competitive performance in durability-focused environments. The substantial improvements in query speed and throughput suggest SurrealDB is a viable alternative for multi-model, transactional workloads, especially where full durability and high concurrency are required. These results could influence deployment decisions in enterprise and high-demand applications.

Background

Previous benchmarks with fsync disabled showed faster results but did not reflect typical production conditions. This latest round emphasizes full durability, aligning with real-world deployment scenarios where data integrity is critical. SurrealDB’s latest internal architecture overhaul—covering query parsing, storage, and indexing—has driven the performance gains. Historically, databases like Postgres and MongoDB have been benchmarks for relational and document models, respectively, but SurrealDB’s multi-model approach and recent optimizations position it as a competitive alternative.

“SurrealDB 3.x’s new query planner and storage engine eliminate per-row decoding overhead, enabling workloads that previously took minutes to complete in seconds.”

— SurrealDB development team

“Running all databases on the same hardware with production-grade configurations provides a fairer, more realistic comparison for enterprise deployments.”

— Benchmarking lead author

What Remains Unclear

While the benchmarks are comprehensive, the results are specific to the tested hardware and configurations. Performance may vary in different environments or with different workload mixes. Additionally, the long-term stability and scalability of SurrealDB 3.x under sustained high load remain to be validated through real-world deployment.

What’s Next

Further testing across diverse workloads and environments is expected, along with ongoing development to close remaining performance gaps with traditional databases like Postgres. SurrealDB’s team plans to release version 3.1, which aims to improve query planner capabilities and index efficiency. Industry adoption and real-world case studies will be critical to assess its practical performance and stability.

Key Questions

How does SurrealDB 3.x compare to Postgres in real-world scenarios?

Benchmarks show SurrealDB 3.x outperforms Postgres in write-heavy workloads under full durability settings, but real-world performance will depend on specific use cases and configurations.

What are the main improvements in SurrealDB 3.x?

The internal overhaul of the query parser, storage engine, and indexing layers has resulted in significant performance gains, especially in full-table scans and CRUD operations.

Is SurrealDB suitable for production environments?

Yes, the benchmarks used production-grade configurations with full durability enabled, indicating its readiness for high-demand, data-critical applications.

Will SurrealDB replace traditional databases?

It offers a compelling alternative, especially for multi-model, transactional workloads, but adoption will depend on further testing, stability, and ecosystem maturity.

Source: Hacker News