Reports & Publications

Ascom Timeplex Integrated Access Node - Data Link Switching (DLSw) - RFC-1434 Compliant Performance - vs. Bay Networks & IBM

Sponsor: Ascom Timeplex
Ascom Timeplex IAD- Data Link Switching (DLSw) - RFC-1434 Compliant Performance vs. Bay & IBM

Abstract

Ascom Timeplex commissioned The Tolly Group to evaluate the performance and functionality of its Integrated Access Node for Data Link Switching, with the main focus on comparing RFC 1434-compliant DLSw performance against competing products from IBM and Bay Networks. The project examined maximum single-session throughput across dedicated T1 links and verified whether the products could keep IBM 3270 SNA sessions alive under heavy WAN congestion, a key requirement for reliable mainframe connectivity over wide-area networks.  


The report explains that traditional Token Ring bridges forward LLC2 traffic but do not protect session integrity when WAN congestion occurs, which can cause session drops and disrupt critical host access. DLSw addresses this by intervening in the previously untouched protocol exchanges of source-route-bridged LLC2 sessions while remaining transparent to the end stations. Tolly tested three products: the Ascom Timeplex Integrated Access Node (IAN) release 4.0.2, the Bay Networks Backbone Concentrator Node 73000 release 7.80, and the IBM 6611 Network Processor models 140 and 170 running MPNP V1 R2. All three were measured using the same methodology.  


Performance testing used live SNA/APPC file transfers across a simulated T1 link at 1.536Mbit/s between two 16Mbit/s Token Ring LANs. The tests ran four times per vendor using SNA request or response unit sizes of 256, 512, 1,024, and 2,048 bytes. According to the throughput table on page 2, Ascom Timeplex delivered 0.70, 1.20, 1.35, and 1.34Mbit/s respectively; Bay Networks delivered 1.10, 1.28, 1.39, and 1.28Mbit/s; and IBM delivered 0.32, 0.64, 1.26, and 1.41Mbit/s. Tolly notes that all of the routers filled more than 80% of the T1 at 1,024-byte and 2,048-byte RU sizes, demonstrating that DLSw could provide WAN throughput comparable to high-end Token Ring bridges.  


Tolly also ran congestion tests over a simulated 56Kbit/s WAN link to determine whether the devices could preserve active 3270 sessions when the line was saturated with SNA or IP traffic. In both test conditions, every product passed: no session was ever lost. For Ascom Timeplex, Tolly had to prioritize TCP/IP and DLSw over other IP traffic during the congestion test, while other products allowed sessions to hang temporarily but recover once congestion subsided. Overall, the report presents the Ascom Timeplex IAN as a capable RFC 1434-compliant DLSw platform that combined strong T1 throughput with stable SNA session survivability under WAN overload conditions.