WebLogic Server provides failover within MAN and WAN networks. This feature provides more reliability by allowing failover to occur across a larger geographic area. It also provides failover across multiple WebLogic Server domains.
To provide failover within a MAN/WAN environment, you must use hardware load balancers. This document outlines the procedures for configuring F5 hardware load balancers to work with WebLogic Server.
For information on configuring WebLogic Server to use MAN/WAN, see Session State Replication Across Clusters. For information on configuring F5 hardware load balancers, see www.F5.com.
The following sections describe how to configure F5 hardware load balancers.
Before performing the procedures described in this appendix, you must have performed the following:
Once these requirements are met, perform the following procedures to configure your load balancers to work within a MAN/WAN environment.
This section describes the procedures for configuring F5 local load balancers to work with WebLogic Server in a MAN/WAN environment.
On each local load balancer you must configure two virtual server IPs as well as a multi-layer pool and a failover trigger pool. The following diagram shows how these pools and virtual server IPs work within a MAN/WAN environment.
In this diagram, multiple managed servers are distributed across separate physical locations. This diagram shows individual managed servers, but this could also represent a clustered configuration as well.
Each local load balancer contains a virtual server IP that references a multilayer pool. The multilayer pool references each of the local WebLogic Server IP addresses and host names and the virtual server of the failover trigger pool. The failover trigger is used to indicate that a site is down. This triggers failover to the other local load balancer.
The following sections describe how to configure multi-layer and failover trigger pools.
Create a new BIG-IP pool on the local load balancer that references each of the local WLS server hostnames and ports to be load-balanced. Then, create a new virtual server that specifies this pool. This virtual server will be utilized by the 3-DNS global load balancer for health monitoring and will later be embedded inside another local load balancer pool/virtual server.
Using the F5 administration utility, create a new BIG-IP pool on the local load balancer that references each of the local WLS server host:ports and also the failover-trigger virtual server. The failover-trigger virtual server must be a lower priority than the WLS Servers. By assigning a lower priority, the failover-trigger virtual server will never receive client requests unless all the WLS servers have failed. Session persistence should be configured also.
A global load balancer type of network hardware that acts as an authoritative DNS server and can distribute web requests across multiple BIG-IP virtual servers based on chosen criteria. Clients send http requests to the global load balancer, which uses built in health monitors to direct the web requests to the optimal server based on the chosen method of load balancing.
The global load balancer must be an authoritative source of DNS because a regular DNS server is incapable of the monitoring that the global load balancer can perform. A regular DNS server would still send http requests to a server that was down if it were next in the default round robin load balancing method. In order to compensate for the multiple shortcomings of a regular DNS server, many vendors (including F5) have created specialized hardware and software that is capable of performing not only DNS resolution but also intelligent routing of network traffic.
The primary steps of configuring an F5 3-DNS global load balancer are: defining its DNS name, configuring the BIG-IP hosts, configuring data centers, and configuring the 3-DNS distribution of work to the virtual servers (VIPs). These are covered in the following sections.
The global server load balancer must be configured to manage its own DNS zone. This is done by creating a new delegation on the local DNS management machine. The following procedures describe how to configure DNS zones.
The 3-DNS global balancer needs to be configured with the addresses of the BIG-IP local load balancers. The following procedures outline how to configure BIG-IP addresses:
In most cases, global load balancers spread service requests to virtual servers in multiple physical sites. These sites are called data centers and you must create two of them. Data centers resolve to the two different subnets of BIG-IP local load balancers.
It is recommended that you configure the 3-DNS device so it will distribute requests evenly to servers in a VIP in one data center. If these servers fail, they should fail requests over to a VIP in the other data center. In order to do this, a wideip address must be created. This wideip address will be the target of client requests, and can be given a fully qualified domain name. The Wide IP defines how connections are distributed to local load balancer virtual servers.
The following procedures describe how to configure wide IPs:
After you have configured your F5 devices, you must configure WebLogic Server to use MAN/WAN failover. For information on configuring WebLogic Server to use MAN/WAN, see Session State Replication Across Clusters in a MAN/WAN.