Load Balancing and Failover - Introduction
Imperva Load Balancer distributes user requests among origin data centers and/or servers in order to achieve optimal performance and response time. In addition, it helps to ensure high availability in the case of a malfunctioning server or data center by routing traffic to a healthy server.
The availability of this feature depends on your subscription. For more information or to upgrade your plan, contact an Imperva sales representative.
- If your site already has one or more load balancers installed, the load balancers’ IP addresses should be entered as server IPs in the Imperva Load Balancing configuration. Imperva will treat each load balancer as if it were a single server.
- Server load balancing using layer 7 distribution algorithms
- Layer 7 (not DNS) based global server load balancing
- Site failover and disaster recovery scenarios
- Health monitoring and server failover
How Does Load Balancing Work?
Imperva’s Load Balancing is based on a network of secure reverse proxies deployed on our globally distributed CDN. Web traffic that is routed through the Imperva network is terminated by those proxies. This allows Imperva to act as a load balancer at the HTTP level by making sure requests are always routed to the origin server with the smallest load, as well as executing geography-based routing decisions at the request level.
Load Balancing at the Request Level
Imperva uses Layer 7 based algorithms to make load balancing decisions at the HTTP request level. The Least Pending HTTP Requests distribution method measures the number of pending HTTP requests for each origin server and sends requests to the origin with the lowest number of pending requests. This method offers a very accurate assessment of the origin servers’ loads and keeps the load evenly distributed among the origin servers.
Global Server Load Balancing (GSLB) at the Request Level
Imperva is quite unique in its use of Layer 7 based algorithms to make GSLB decisions at the HTTP request level (as opposed to DNS-based GSLB). Layer 7 GSLB allows for quick (non TTL-reliant) responses to server and data center malfunctions.
Site Failover and Disaster Recovery (DR) Scenarios
The Imperva Load Balancer can also play a major role in DR planning, acting as an automated solution for site failover .
By using the health monitoring feature, Imperva immediately detects that the primary site is down and automatically fails over to the standby site.
Health Monitoring and Server Failover
Imperva supports advanced health monitoring, constantly checking the origin servers to detect malfunctions and allow immediate server/site failover.
Customers have complete control over the monitoring system. They can calibrate its sensitivity, configure specific URLs to be monitored and define the exact responses that are expected to be received.