Abstract

IBM commissioned The Tolly Group, as part of its 1994 Industry Benchmark for Local Token Ring Bridges, to evaluate the IBM LANStreamer Token-Ring Network Bridge/DOS with the main focus on measuring local Token Ring bridge performance across source-route bridging scenarios. The report emphasizes frame-handling efficiency, effective unidirectional throughput, and the bridge’s ability to deliver near-wire-speed performance on larger frame sizes in 16Mbit/s Token Ring environments.  

The December 1994 Technology Spotlight describes the IBM LANStreamer Token-Ring Network Bridge/DOS as a PC-based bridge that runs on an IBM-compatible system using two IBM LANStreamer Token Ring adapter cards. The tested configuration ran on an IBM PS/2 77 with a 486 DX2 66MHz processor and used IBM’s LANStreamer technology for Token Ring connectivity. According to the product-specification panel on page 3, the bridge software version was 1.11, it supported two fully interconnected interfaces per chassis, 4 or 16Mbit/s Token Ring operation over STP and UTP media, Source Route bridging, and frame sizes up to 8,192 bytes. Notable features listed include LAN Network Manager support and an extended source-route hop count limit of 14.  

Tolly’s benchmark suite measured overall throughput for source-route bridged frames using seven frame sizes: 28, 64, 256, 512, 1,024, 2,048, and 4,096 bytes. As shown in the chart on page 1, the IBM LANStreamer bridge delivered 4.19Mbit/s at 28-byte frames, 8.55Mbit/s at 64-byte frames, 14.58Mbit/s at 256-byte frames, 15.00Mbit/s at 512-byte frames, 15.48Mbit/s at 1,024-byte frames, 15.73Mbit/s at 2,048-byte frames, and 15.86Mbit/s at 4,096-byte frames. Tolly notes that this corresponds to 18,686fps at 28-byte frames, 16,700fps at 64-byte frames, 7,120fps at 256-byte frames, 3,663fps at 512-byte frames, 1,890fps at 1,024-byte frames, 960fps at 2,048-byte frames, and 484fps at 4,096-byte frames. Near-wire-speed throughput, defined by Tolly as 90% or more of available bandwidth, was achieved for 256-byte frames and larger.  

The report explains that smaller frames produce lower effective throughput because Token Ring bridges forward traffic on a frame-by-frame basis and are constrained by frame-processing limits in the adapter chipset. Even so, the IBM LANStreamer Token-Ring Network Bridge/DOS delivered strong results and reflected the broader improvement in bridge performance that Tolly observed in its 1994 Local Token Ring Bridge benchmark. Overall, the report presents the IBM LANStreamer bridge as an inexpensive, PC-based Token Ring bridging solution that combined high frame-processing capability with near-wire-speed throughput for larger source-route bridged frames.