<mode> | Choose one of the following modes: • refl-peer - Automatically estimate the TCP bandwidth to control congestion if the peer SteelHead appliance is also estimating bandwidth. This setting makes satellite optimization easier to configure. We recommend this setting on the server-side SteelHead in a satellite network. For example, suppose you have a large number of remote SteelHeads communicating with a server-side SteelHead. Rather than defining an in-path rule for every subnet that communicates with a remote SteelHead over a satellite link, it is easier to enable the global always setting on the remote SteelHead and this refl-peer setting on the server-side SteelHead. The server-side SteelHead can then detect the remote SteelHead during the connection setup and communicate with it over the satellite network. When this setting is enabled on both SteelHeads, TCP bandwidth estimation does not occur. At least one peer SteelHead must be set to always to estimate TCP bandwidth. Enabling this option requires an optimization service restart. • always - Always estimate the TCP bandwidth to control congestion globally on all traffic sent by this SteelHead appliance, regardless of the setting on the peer SteelHead. Enabling this option also communicates this configuration to the peer SteelHead appliance so the peer can use TCP bandwidth estimation when it sends traffic to this SteelHead appliance. We recommend this setting on client-side and server-side SteelHeads in a satellite network. Enabling this option requires an optimization service restart. • disable - Disables bandwidth estimation mode. If this option is used, the TCP congestion control mode is set back to the default, which is standard TCP optimization. |