Introducing adb: Arize’s AI-native datastore

Arize has made real-time ingestion available across all Arize AX workspaces and now processes terabytes of AI observability data daily through adb, its custom analytical database designed for high-volume generative AI workloads and rapid ad hoc analytics. Built because conventional databases did not meet its scale, update-frequency, export, and durability requirements, adb separates stateless compute from object-storage-based data, uses Parquet, Iceberg, Arrow, and Apache Flight for standards-based storage and direct access, and aims to provide sub-second queries across billions of rows. Its architecture supports continuous streaming ingestion through an in-memory query layer, direct batch uploads to blob storage, indexing for unified querying, configurable compaction for varied data sizes, and cache management intended to reduce costs. Originally derived from Apache Druid but substantially rewritten, adb reflects Arize’s preference for an “anti-monolith” model in which open data formats in durable storage can be served by flexible, lightweight query engines.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.