Visara LINCS Features - Manual (Page 18)

8
LINCS Features
707021-001
which is more efficient than algorithms that retransmit the missing packet and all following
ones. This selective retransmit algorithm preserves packet order, and is the foundation for the
Multiple Link Transmission Group (MLTG) support in HPR. RTP handles link failures by
computing ANR labels for a new path that meets the class-of service requirements, and switching
to it without notifying or disrupting higher protocol layers.
RTP establishes RTP connections to carry session traffic across a HPR subnet. These
connections are “transport pipes” that connect two HPR nodes over a specific path in a HPR
subnet. Multiple sessions may share the same RTP connection if they are using the same class
of service. Intermediate HPR nodes have no knowledge of the RTP connection. They simply
route the session traffic based on the ANR routing information.
RTP performs the following functions:
• Segmentation to the size needed for the smallest link
• Reassembly of segments Fast RTP connection setup and dissolution
• Selective retransmissions – RTP retransmits only missed or corrupted packets, instead of
every packet since the error occurred.
• Sequence checking, in-order delivery
• End-to-end error recovery – HPR takes advantage of high-speed links. Since high-speed
links are more reliable, they do not need the level of error recovery found in base APPN.
RTP performs error recovery on an end-to-end basis, instead of requiring link-level error
recovery on each intermediate link (as in base APPN). By only checking for errors at the
endpoints of a RTP connection, the number of flows required for error recovery is reduced.
• Nondisruptive path switching – If a link or node goes down, RTP automatically reroutes the
data without disrupting the traffic flow. RTP connections are reestablished over a new route
that bypasses the failed link or node. Missed data is automatically recovered using end-to-
end error recovery.
Adaptive rate-based flow/congestion control (ARB)
Base APPN performs adaptive session pacing at each node in the network. This method of
flow control works well in networks with various link types operating at different speeds.
Networks with high-speed links, however, can reduce the amount of processing done at each
node by using adaptive rate-based congestion control (ARB) at the RTP endpoints. ARB
attempts to predict when congestion will occur and reduce a node’s sending rate before this
happens. Each node samples the rate at which it receives and sends packets. When buffering
limits are approached, ARB appends rate messages to data packets telling the end nodes to
speed up or slow down, instead of waiting until after congestion develops and packets are
discarded, requiring retransmission. ARB avoids congestion, instead of reacting to congestion,
which yields higher link utilization.
HPR Configuration and Management
Configuration – HPR is configured for an APPN circuit by setting the APPN HPR field in
the circuits Link Profile to the desired value.
Management – The APPN Menu in Central Control Mode contains the following utilities
which have HPR specific information.