Abstract

Barr Systems commissioned Tolly evaluate the DATMODE=FULL  performance of its T1-SYNC SDLC WAN adapter. Tests will show the benefits of full duplex WAN transmission vs. half duplex WAN transmission in a Microsoft Windows NT environment for 3270 mainframe and APPC file transfer traffic. 

This Tolly Group test summary evaluates the Barr Systems T1-SYNC SDLC adapter for Microsoft Windows NT and focuses on the performance impact of IBM’s DATMODE=FULL enhancement. Tested in January 1994, the adapter was measured under Microsoft Windows NT 3.1 and SNA Server 2.0 to compare conventional DATMODE=HALF operation with full-duplex DATMODE=FULL behavior. The report’s central conclusion is that DATMODE=FULL can provide a substantial throughput improvement over traditional SDLC operation, in some cases effectively doubling performance without increasing line speed. Barr positioned the T1-SYNC as a low-cost “upgrade-in-place” option for organizations facing poor SDLC throughput. 

The report explains that DATMODE=FULL is an IBM enhancement to SDLC that allows session data to flow in both directions simultaneously. In contrast, DATMODE=HALF limits transmission to one direction at a time, cutting the effective potential throughput of the data link. Tolly emphasizes that the observed performance gain comes from the DATMODE=FULL protocol enhancement itself, not simply from using full-duplex communications hardware. According to the chart on page 1, the adapter showed percentage gains from DATMODE=HALF to DATMODE=FULL ranging from 64% to 111% at 9.6Kbit/s, 79% to 100% at 19.2Kbit/s, and 97% to 114% at 56Kbit/s across selected SDTF test suites. 

Tolly also ran bidirectional APPC file-transfer tests between two Windows NT systems to stress the adapter with multiple simultaneous transfers. As shown in the chart on page 2, throughput improved by 73.79% at 19.2Kbit/s, 95.75% at 56Kbit/s, and 95.04% at 128Kbit/s when switching from DATMODE=HALF to DATMODE=FULL. The report states that when the card was set to DATMODE=FULL, throughput in several cases more than doubled, and every lab test showed improvement. 

The test environment used a Gateway 2000 486DX2-66 with 32MB of RAM, Barr’s T1-SYNC adapter, and an IBM 16/4 token-ring card running Microsoft Windows NT 3.1 and SNA Server 2.0. The system was configured as a physical unit with four logical units, while three Compaq Deskpro 4/33i systems acted as Windows for Workgroups-based 3270 emulator clients. The SDLC link connected through a CR Systems SNA Xtender, an SDLC-to-LLC converter, to an Applied Computer Technology SNA Development and Test Facility mainframe emulator. The diagram on page 4 illustrates this SDLC test bed. Barr’s proprietary file-transfer software created three simultaneous transfers in each direction, and actual frame timestamps were checked for verification. 

Overall, the report presents the Barr Systems T1-SYNC as a practical way to unlock the performance benefits of DATMODE=FULL in Windows NT SNA environments. At the time of testing, Tolly notes that Barr’s product was the only PC SDLC adapter supporting DATMODE=FULL, replacing Microsoft’s native DATMODE=HALF-only link behavior and offering users the potential for major SDLC throughput gains through hardware and software upgrades rather than higher-speed lines alone.