Configuring Network Integration Features : About deployment scenarios : About deployment design
  
About deployment design
You can use multiple SteelHead Interceptors to ensure that optimization occurs even when there are multiple, physically separated paths to the WAN, or to ensure that optimization continues during a planned or unplanned SteelHead Interceptor outage. Use multiple SteelHeads to ensure redundancy or to specify that certain applications, subnets, and hosts are optimized by different SteelHeads.
For details on deployment scenarios, see the SteelHead Interceptor Deployment Guide.
About LAN-side versus WAN-side SteelHead
You connect SteelHeads to the local network on either the LAN side or WAN side of the SteelHead Interceptor. For most deployments, LAN-side SteelHead placement minimizes the amount of traffic that traverses the SteelHead Interceptor. LAN-side SteelHead placement is, therefore, the recommended choice. You can’t cable the SteelHead directly to the SteelHead Interceptor.
About Layer-2 versus Layer-3 connectivity
The SteelHead and SteelHead Interceptor in-path IP addresses can be on the same, or different subnets. The actions taken to redirect traffic are the same during autodiscovery and optimization. Unlike when you use WCCP to redirect traffic to a SteelHead, there is no resource use or MTU concerns from having the SteelHead and SteelHead Interceptor in-path IP addresses on different subnets.
About multiple SteelHead link support
A SteelHead can have multiple in-path interfaces configured so that SteelHead Interceptor can use multiple IP addresses to reach the same SteelHead. The SteelHead Interceptor only redirects traffic to one interface on the SteelHead at any time.
About multiple SteelHead support
Using the steelhead communication commands, the clustered SteelHeads configure each clustered SteelHead Interceptor as a connection forwarding neighbor. Each clustered SteelHead typically has one SteelHead Interceptor in-path interface (for example, inpath0_0, but this varies by model) configured as the clustered SteelHead main interface, and the other SteelHead Interceptor interfaces configured as additional choices. Don’t configure a connection between the clustered SteelHeads.
You can use the RiOS data store synchronization between any local pair of SteelHeads, including SteelHead pairs in a SteelHead Interceptor cluster. You typically perform data synchronization when the load-balance configuration on the SteelHead Interceptor causes similar traffic to be sent to the pair of SteelHeads.