As described previously, Instant Connect Enterprise is architected around the principle that as little audio processing be done in the network as absolutely needed. However, in the case of patches, a design decision was made to centralize patching so that a patch could outlive the administrator or device that created it.

Recall that ICE clients, and not a centralized media server, are responsible for mixing the audio of simultaneously received audio together. Furthermore, clients are able to receive and decode audio streams comprised of heterogenous CODECs, too (that is, each talker on a channel could transmit using a different CODEC). As a benefit of these capabilities, the patch server does not need to decode or mix audio streams together. Instead, when processing unencrypted audio streams, the patch server joins each channel participating in the patch and simply retransmits audio packets received on one channel to all the other patched channels. For encrypted channels, the process is slightly more CPU-intensive: the patch server must decrypt and re-encrypt traffic as it gets mirrored on participating channels.

In order to use patching functionality, channels should have ICE Rallypoint enabled. The patch server functions like a client in that it connects to Rallypoints to receive and transmit patched audio. This implies that the patch server must have reachability to any Rallypoint used by a channel.