Browse Prior Art Database

Sliding Sub-Window Protocol in the Systems Network Architecture

IP.com Disclosure Number: IPCOM000042777D
Original Publication Date: 1984-Jun-01
Included in the Prior Art Database: 2005-Feb-04
Document File: 2 page(s) / 14K

Publishing Venue

IBM

Related People

Bharath-Kumar, K: AUTHOR [+2]

Abstract

This article describes a protocol for resequencing in the Systems Network Architecture (SNA) Virtual Route (VR) that efficiently integrates with the existing window flow control protocol. In SNA, message streams of a certain session may go out of sequence while traversing Transmission Groups (TGs) that contain multiple physical links. Today, the resequencing of messages is carried out at the TG receiver. However, instead of resequencing messages at every hop in the network, it is substantially more efficient, in terms of the response time/throughput seen by end users, to resequence once at the VR receiver [*]. However, this reassignment of function may cause increased buffer requirements at the VR receiver as follows.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 51% of the total text.

Page 1 of 2

Sliding Sub-Window Protocol in the Systems Network Architecture

This article describes a protocol for resequencing in the Systems Network Architecture (SNA) Virtual Route (VR) that efficiently integrates with the existing window flow control protocol. In SNA, message streams of a certain session may go out of sequence while traversing Transmission Groups (TGs) that contain multiple physical links. Today, the resequencing of messages is carried out at the TG receiver. However, instead of resequencing messages at every hop in the network, it is substantially more efficient, in terms of the response time/throughput seen by end users, to resequence once at the VR receiver [*]. However, this reassignment of function may cause increased buffer requirements at the VR receiver as follows. In the worst case, for existing SNA, the VR receiver is committed to 2W-1 (where W is the flow control pacing window size) additional messages when deciding to send a positive pacing response (VRPRS). This represents the maximum outstanding, i.e., authorized through pacing responses but not yet received. Introducing VR resequencing (also referred to as REFIFO) may increase this worst case for the VR receiver since some messages of previous pacing windows may be still held at the receiver for resequencing when a pacing response is sent. This assumes that the maximum sequence number or the resequencing window size (M) is large enough. The worst case buffer requirement is (X+1)W-1, where X is the largest integer such that this worst case value is less than M. By choosing M such that X=1 obviously the buffer requirement is kept the same as before. However, with this small M, the resequencing window protocol that does not permit wrap around, like the one used in TG resequencing, is inefficient. This is because the entire VR pipe has to emptied out of old sequence numbered messages before starting the new window of sequence numbers. Hence, a more efficient resequencing window protocol is needed. Let us consider the protocol that uses M=2W and entails a simple addition to the VR window protocol. The protocol consists of withholding the pacing response at the VR receiver till all the messages of the previous pacing window have arrived and are in sequence. The pacing response arriving ...