Additional copy of a search service's engine and indexes. Replicas spread queries for load balancing and improve availability, so adding them is the fix when queries are throttled with HTTP 503.
Also called replica.
Read more: Microsoft Learn
In the Ultra Transcenders books
Each book explains Azure AI Search replica in context, with comparison tables and the common traps.
Terms in this definition
Related terms
- Always On availability group
High-availability feature of SQL Server that keeps synchronous or asynchronous replicas on different servers or VMs without shared storage. As every replica is a live SQL Server, it is an expensive way to provide DR.
- Data persistence
An Azure Managed Redis option that keeps a copy of the cache on a managed disk, as either an AOF log or RDB snapshots. That lets the same cache be restored after both its primary and replica go down. It is no substitute for backup, and active geo-replication rules it out.
- Database scoped configuration
Per-database options, for example MAXDOP, IDENTITY_CACHE and scalar UDF inlining, that are set on an individual database instead of server-wide; a secondary replica can be given its own value.
- Enhanced Log
Introduced for Storage Replica in Windows Server 2025, this log type is faster than, and supersedes, the Traditional log.
- GC
Serves forest-wide searches and logons on port 3268 by holding a partial replica of every object. Where no global catalog exists at a site, universal group membership caching can fill in.
- General Purpose
An Azure SQL service tier that pairs a single compute node with remote storage. Compared with Business Critical it has higher latency and lacks a synchronous replica.
- Mirrored database
What mirroring creates in Fabric: an almost real-time replica of a database held elsewhere, stored in OneLake as Delta tables. Users can query it with T-SQL through the read-only SQL analytics endpoint that comes with it.
- Multi-region writes
Cosmos DB option under which each region holds a replica that accepts writes.