Query Load Balancing - Performance Boost for BI Applications

For business intelligence (BI) applications, heavy calculations make query load balancing crucial in effective use of SQL Server resources. Complex BI applications need sophisticated manipulation of database objects before generating reports. These are not pure read-only applications. "Sticky" load distribution algorithm is required to ensure the correct operations.

Network load balancers are designed to distribute stateless web traffic. They typically use cookies to maintain client session states. They lack continuous session management capabilities. This affects the effectiveness of their load balancing algorithms and can require application code to directly monitor individual SQL Server states.

DB^x™ load balancer is a by-product of DB^x™ SQL cluster engine. It manages all concurrent client-SQL Server sessions. It can distribute query load based on sessions or individual queries. It responds immediately to all SQL Server state changes. Applications with proper re-do error-handling logic are automatically SQL Server and DB^x™ fault tolerant.

DB^x™ failures are protected by built-in IP-takeover function for LAN deployments and by DNS for WAN deployments.

A unique feature in DB^x™ cluster is that it allows adding SQL Servers using an automatic resynchronization tool without shutting down service. This also allows individual SQL Servers to be taken offline for repair and resynchronized without affecting cluster service status.

In the DB^x™ load balancer, there are four load-balancing heuristics: round robin, least server load, least waiting query and least waiting connections. Adjusting the "performance coefficient" for individual SQL Servers can further customize each of these algorithms.