Weaviate v1.40.0: Namespaces GA, Drop Vector Index, Dedup Backups, HFresh & Search REST API
Weaviate version 1.40.0 delivers general availability for Namespaces offering multi-tenant isolation, introduces dedicated REST search endpoints, supports dropping vector and inverted indices for disk space savings, includes numerous core performance improvements, and brings new backup and integration capabilities.
What changed?
Weaviate v1.40.0 brings several notable infrastructure updates: - Namespaces are now generally available, providing control-plane and data isolation for multi-tenant clusters. This feature requires a Weaviate license. - Dedicated REST API endpoints for search are introduced and enabled by default. This covers hybrid search, near-object, near-vector, aggregate counts, and structured reference selection, removing the previous experimental flag. - Support for dropping vector indices and inverted indices from existing properties is now generally available, allowing administrators to reclaim disk space. This is RBAC and multi-tenancy integrated. - New preview functionality lets properties be reindexed with a different index type. - Numerous performance optimizations have been delivered for HFresh, including reduced memory usage and disk writes. HFresh MUVERA now decouples routing and rerank budgets. - Opt-in deduplicated backups allow archiving data once for all replicas, improving backup efficiency (requires license). - Additional improvements include async replication hardening, enhancements to backup reliability, new generative integration modules (DigitalOcean, Meta AI), and usage metrics updates.
Why does it matter to an everyday developer?
These changes offer practical benefits for developers managing or building on Weaviate instances: - Dedicated REST search endpoints enable straightforward, robust integration into applications needing semantic, hybrid, and aggregate search. Default enablement removes migration friction. - Namespace isolation allows for stricter separation between tenants, supporting multi-user SaaS scenarios on shared infrastructure and improving data security and operational control. - The ability to drop no-longer-needed indices helps keep storage requirements and operational costs under control without affecting active workloads. - Improved backup strategies (deduplication) and reliability mean reduced recovery times and safer protection in multi-replica setups. - Core index and vector optimizations (HFresh) and async replication hardening improve overall cluster throughput and resilience.
