High Availability

In High Availability (HA) configurations, the redundant Silver Peak appliances are deployed in router mode and either WCCP or PBR redirects flows from the routers to the appliances.

The redundant appliances may be configured Active/Active or Active/Backup. This is determined by how the WCCP or PBR redirection is configured on the routers and the appliances.

For the purposes of discussion, we’ll assume that HA is configured in the same location as the servers and we’ll refer to the HA (redundant) appliances as “server-side”. We’ll refer to the non-redundant appliances as “client-side”. Of course, it doesn’t need to be this way — it’s possible to have redundant Silver Peak appliances in offices without servers.

Auto-optimization or Explicit Route Maps?

In HA configurations, the decision about whether to use auto-optimization or explicit route maps has further implications. Considerations include the following:

•

The network may already have inherent asymmetry, relative to the deployment you want to configure.

•

Provisioning redundant appliances may introduce network asymmetry where none existed before.

•

Depending on exactly how a router’s inbound and outbound redirection statements are configured, it’s possible to arrive at an asymmetric condition.

•

With load sharing (Active/Active) configurations, asymmetry is a fact of life.

Asymmetry Mitigation

Flow redirection can prevent TCP asymmetry in high availability environments.

For the appliances, this requires configuring HA (or redundant) appliances as peers, and enabling flow redirection. Both tasks are on the Configuration - Flow Redirection screen.

Where it’s an element of any deployment chapter in this guide, the instructions include the configuration steps.

High Availability with Explicit Route-Maps

When auto-optimization is not enabled, explicit route maps in the appliance determine how to route traffic into the tunnels for optimization.

We’ll examine two high availability situations from the point of view of the client-side appliance:

Asymmetry in Active/Backup Configurations

One tunnel carries all the traffic. If that link goes down, then the Backup appliance receives the client’s traffic via another tunnel. Enabling flow redirection on the peered server-side appliances ensures that the same tunnel carries those flows back to the client.

Asymmetry in Active/Active Configurations

The server-side router is load balancing and determines which peer appliance receives the returning flow. Enabling flow redirection among peers prevents TCP asymmetry.