The ISA RET server is supported within an IES or VPRN service context as applicable to the platform.
Whether the RET server is active for a specific multicast channel is defined in the multicast information policy where channels are defined. The channel configuration for the RET server within the policy is an explicit enable/disable of the local RET server (that is, whether the channel should be buffered), the RET buffer size for the channel in the ISA and a channel type (Picture-in-Picture (PIP), Standard Definition (SD) or High Definition (HD)). The RET buffer should be large enough to account for the round trip delay in the network; typically, a few hundred milliseconds is sufficient.
In an IES or VPRN service, up to 16 IP addresses can be assigned to a video interface.
The video policy within the multicast information policy defines the characteristics for the how the RET server should respond to NACKs received on an IP address. The different characteristics defined in a RET server ‟profile” are for each channel type (PIP, SD and HD):
enable/disable for the RET server (that is, whether requests should be serviced or dropped)
the RET rate (as a percentage of the nominal channel rate)
Typically, RET replies are sent below line rate because most dropped packets occur in the last mile and sending RET replies at a high rate may compound any last mile drop issues.
The IP address(es) of the RET server are defined in the unicast service instance, whereas the UDP port for the RET server is defined in the ‟default” bundle in the multicast information policy. The same UDP port is used for all RET server IP addresses that use the particular multicast information policy.
The ISA RET server supports the network model where there are separate service instances for unicast and multicast traffic that are cross-connected and multicast replicated downstream in the network. If there are separate multicast and unicast service instances, the unicast and multicast services must use the same multicast information policy.