SurrealDB 3.x Benchmarks Against Incumbents, Highlighting Fsync Impact

This review analyzes SurrealDB's official performance claims for its 3.x release, comparing it against established databases like Postgres, MongoDB, Neo4j, and Redis, with a focus on Fsync considerations.

The Answer Up Front

SurrealDB 3.x is positioned as a compelling option for founders building applications that require a flexible, multi-model database capable of handling diverse data types—graph, document, key-value, and relational—within a single system. Its claimed performance, particularly with fsync enabled, suggests it could simplify infrastructure by consolidating multiple specialized databases. Skip SurrealDB if your primary concern is long-term operational maturity or if your workload is strictly confined to a single, well-understood data model where highly optimized, single-purpose databases already excel. The bottom line: SurrealDB 3.x presents an ambitious, potentially powerful converged database, but its performance claims require independent validation for mission-critical deployments.

Methodology

This v0 review draws on the founder's published claims at https://surrealdb.com/blog/surrealdb-3-x-by-the-numbers, accessed on 2026-06-01. The blog post, authored by itsezc, details performance benchmarks for SurrealDB 3.x (released May 29, 2026) against a suite of established databases: PostgreSQL 16.3, MongoDB 7.0.8, Neo4j 5.17.0, and Redis 7.2.4. Benchmarks were conducted across various operations including inserts, reads, updates, and deletes, with a particular emphasis on the impact of fsync (or sync_commit in Postgres) on write performance and durability. The source provides specific commands and configurations used for each database, allowing for a degree of reproducibility. What's covered in this review are the founder's own claims regarding performance numbers, the described testing environment, and the technical rationale behind their methodology. What's NOT covered are independent performance benchmarks, long-term workflow integration, operational overhead in production environments, or edge-case performance scenarios. Update cadence: re-tested when claims diverge from observed behavior.

What It Does

A Converged Database Vision

SurrealDB aims to be a single, converged database for modern applications, integrating capabilities typically found in separate systems. It supports graph, document, key-value, and relational data models, allowing developers to interact with data using a SQL-like query language (SurrealQL). The 3.x release focuses heavily on performance optimizations, particularly in its storage engine, aiming to deliver competitive speeds across these diverse workloads.

Benchmarking Suite and Fsync Focus

The core of the 3.x announcement is a detailed benchmark comparison. The founder claims SurrealDB 3.x outperforms or matches several specialized databases in their respective domains. For instance, in fsync enabled write operations, SurrealDB 3.x claims to achieve 15,000 writes per second, significantly higher than PostgreSQL's reported 2,000 writes per second and MongoDB's 1,000 writes per second in similar configurations. The blog post explicitly highlights the importance of fsync for data durability, arguing that many benchmarks omit this critical setting, leading to misleading performance figures. SurrealDB's internal architecture is designed to optimize fsync calls, which the founder attributes to its strong performance in these durability-focused tests.

Specific Performance Claims

Across various synthetic benchmarks, SurrealDB 3.x claims to demonstrate strong performance. For graph operations, it reportedly offers faster pathfinding and relationship traversal than Neo4j in certain scenarios. For key-value workloads, it claims to approach Redis's speed for basic operations while offering more complex data structures. The founder reports that the new storage engine in 3.x is a key factor in these improvements, allowing for more efficient data handling and reduced overhead across different data models.

What's Interesting / What's Not

What's interesting is SurrealDB's explicit focus on fsync in its benchmarks. Many database performance comparisons either ignore this crucial durability setting or relegate it to an afterthought. By making fsync-enabled performance a central pillar of their 3.x claims, SurrealDB addresses a common blind spot in synthetic benchmarks, which often prioritize raw throughput over data integrity. The ambition to be a high-performance, multi-model database that can genuinely compete with best-of-breed solutions in their specific niches is also noteworthy. If these claims hold up under independent scrutiny, SurrealDB could significantly simplify database architectures for many startups, reducing the need for polyglot persistence and its associated operational complexity.

What's not interesting, or rather, what requires significant caveats, is that all performance numbers are founder-provided. The

The investor read

The database market continues its trend towards convergence, with tools like SurrealDB attempting to capture multiple workloads previously served by specialized databases. This signals a desire from developers for simpler stacks and reduced operational overhead. While the 'jack of all trades' approach is appealing, the challenge lies in truly outperforming or matching incumbents across diverse benchmarks, especially when considering factors beyond raw speed like ecosystem maturity, tooling, and community support. For investors, SurrealDB's viability hinges on independent verification of its performance claims and its ability to build a robust ecosystem. Its focus on fsync is a smart tactical move to differentiate, but the long-term play requires proving it can handle real-world scale and complexity as effectively as its specialized competitors. A successful outcome would make it a strong acquisition target for larger cloud providers or enterprise software companies seeking a modern, converged data platform.

Pull quote: “By making fsync-enabled performance a central pillar of their 3.x claims, SurrealDB addresses a common blind spot in synthetic benchmarks, which often prioritize raw throughput over data integrity.”