WCCP load distribution
The WCCP protocol provides the assignment method for dynamic symmetric and asymmetric load distribution in a cluster. WCCP detects node failures and performs redistribution based on the configuration communicated to it by Content Gateway.
- Load distribution is configured in Content Gateway Manager and is pushed to the WCCP devices.
- Load distribution is configured per service group. For each service group:
- Participating cluster members must be registered to the service group. (The WCCP device makes no decisions about load balancing.)
- The HASH or MASK assignment method is selected. HASH is typically used with the GRE forward/return method, and MASK with the L2 forward/return method.Important: MASK was developed specifically for the Cisco Catalyst series switches, and is one of the key characteristics that enable WCCP interception to be performed completely in hardware on these platforms. It should be used only with devices for which there is documented support.
- One or more distribution attributes are selected. Typically the destination IP address is used.
- If load is to be distributed to different cluster members in different proportions, a weight value is set on each cluster member. These values determine the proportion of requests each
will receive relative to other members of the cluster. This option is only useful if the Synchronize in the Cluster option is disabled. See Configuring service groups in the Content
Gateway manager.
Asymmetric load distribution using the weight value is helpful when:
- There are multiple Content Gateway servers with different performance capabilities.
- The Internet traffic profile doesn’t lend itself to even distribution due to preferences for specific origin servers (and therefore destination IP addresses).