November 2026 Summaries

SurrealDB 4 and its Cloudflare-based serverless platform, SurrealDB Earth, launched on 3 November 2026, offering isolated databases for individual agents, users, or customers that activate on demand, scale to zero when idle, and are billed by requests and storage. Earth is available through free SurrealDB Cloud accounts and is designed for applications ranging from multitenant SaaS products to AI agents with persistent memory. SurrealDB 4 supports combined vector search, full-text search, structured filters, and graph traversal within a single transactional query, enabling agents to retrieve memories by semantic similarity, keywords, metadata, and relationships. It also adds PostgreSQL and MySQL wire-protocol support for defined subsets of each dialect, although several capabilities such as recursive CTEs, triggers, roles, certain vector features, and stored procedures remain unsupported. Planned work includes database branching, faster cold starts, and programmatic database creation, while the product is source available under the Business Source License 1.1, which restricts offering it as a third-party database service until its planned conversion to Apache License 2.0 in 2030.

SurrealDB can support AI agent memory by combining vector similarity search, full-text search, document-field filtering, and graph traversal in a single SurrealQL query and transaction. The example models memories with text, embeddings, speakers, timestamps, and types; entities such as people, systems, and projects; and relationship edges connecting memories to entities and entities to one another. HNSW indexes retrieve semantically similar memories, BM25 full-text indexes find exact or stemmed word matches, and graph queries reveal related systems, projects, and owners, while transactions provide consistent reads and make newly written memories immediately searchable. The post also demonstrates that uniqueness constraints can roll back an entire memory write when duplicate entity links are detected, preserving data consistency. It concludes that while developers can build custom agent-memory systems with these primitives, SurrealDB Agent Memory provides managed ingestion, entity and relationship extraction, fact reconciliation and history, tri-temporal tracking, hybrid ranking, and user or team scoping.

SurrealDB 4 has reached general availability alongside SurrealDB Earth, a serverless Cloudflare-based offering designed to provide isolated, automatically scaling database instances for large numbers of users and AI agents. The release introduces a rebuilt query engine, SurrealKV 1.0 storage engine replacing RocksDB, more than 240 fixes, and a test suite of over 4,500 SurrealQL files intended to improve production reliability and performance. SurrealDB now supports PostgreSQL and MySQL wire protocols for defined subsets of those SQL dialects while retaining SurrealQL and ISO GQL graph-query capabilities, enabling clients to use familiar SQL tools alongside multi-model document, graph, text, and vector-oriented queries. Its Agent Memory service is positioned as a managed way for agents to retrieve semantic similarities, relationships, and source content within a transaction, while SurrealDS provides distributed durable storage across disks, zones, and object storage for cloud multi-node and enterprise deployments. New deployments are encouraged to use version 4, whereas existing 3.x users are advised to wait for 4.1 for production upgrades or cautiously test the experimental one-way in-place migration process, with further SQL functions, database branching, forking, and rapid copies planned next.

SurrealDB Earth is a serverless database offering built on Cloudflare Durable Objects and R2 that can be deployed through SurrealDB Studio or the `surreal` command-line interface, with instance creation and cold starts designed to complete in milliseconds. Users need a SurrealDB Cloud account and can select the free or Pro plan, configure immutable data residency in the EU, US, or worldwide, optionally provide a preferred regional location, and set controls such as cells and IP network policies. The CLI’s cloud component supports authentication, organization selection, unattended deployment with default settings, interactive deployment, and creating an instance while immediately opening a SurrealQL shell to run queries. Existing instances can also be queried, exported, listed, or deleted through CLI commands, while JSON output supports scripting and automation.

SurrealDB 4 is a generally available database release aimed at supporting AI agents and applications that need durable memory, transactional state, graph relationships, documents, vector search, and full-text retrieval in a unified system. The release introduces a rebuilt query core, over 240 fixes, PostgreSQL and MySQL wire-protocol support with defined SQL subsets, MongoDB-driver compatibility through mql.js 1.0, and additional features including ISO GQL, schema migrations, permission-aware vector search, and a Python SDK. It is accompanied by SurrealDB Earth, a Cloudflare-based serverless service that creates isolated, transactional databases on demand and scales them to zero when idle, and SurrealDS, an object-storage-backed distributed architecture for multi-node transactional deployments that separates compute from durable storage. The company presents the platform as a way for workloads to grow from individual agent databases to large shared clusters while retaining the same data model. Although SurrealDB 4 is recommended for new deployments, existing 3.x users are advised to wait for version 4.1 before production migrations, while an experimental one-way in-place upgrade process is available with backups, compatibility checks, storage conversions, potential index rebuilding, and removal of SurrealML models.

SurrealDS is a new distributed storage engine introduced alongside SurrealDB 4 for multi-node, fault-tolerant deployments across availability zones, available through SurrealDB Cloud clusters and Enterprise Edition. It separates compute from durable storage by keeping active data, caches, and commit logs on database nodes while storing persistent database pages in shared object stores such as Amazon S3, Azure Blob Storage, and Google Cloud Storage. The architecture is intended to scale compute and storage independently, use ordered commit logs and multi-zone write durability to preserve consistency, and allow nodes to serve validated reads without routing all traffic through a single leader. Changes are periodically consolidated into copy-on-write, content-addressed storage pages, reducing duplicated data and enabling recovery or node replacement without relying on each node to maintain a complete dataset. Designed for Kubernetes deployments, SurrealDS includes operator-managed placement, networking, upgrades, and health controls, while its distributed protocols and recovery behavior are tested through simulations, formal models, fault injection, and performance testing.