Abstract

FDDI Adapters: A Sure Cure for the Bandwidth Blues

This July 1994 Data Communications lab test, conducted by Kevin Tolly of The Tolly Group and David Newman of Data Communications, evaluated FDDI network adapters designed to bring 100Mbit/s backbone-class bandwidth directly to high-performance LAN servers and workstations. The article examined whether FDDI adapters could relieve congestion caused by heavily used Ethernet and token-ring LANs, especially in environments where servers were burdened by large file transfers, multiple clients, and rising demand for networked applications.

The lab focused primarily on EISA-based FDDI adapters for PC servers and also tested two SBus adapters for Sun workstations. Vendors and products tested included Cisco Systems, IBM, Madge Networks, Network Peripherals, SysKonnect, and 3Com in the EISA category, plus Cisco Systems and Network Peripherals in the SBus category. The test environment included Novell NetWare servers, Ethernet and token-ring client LANs, and Sun workstation configurations. Testing measured throughput across multiple frame sizes and client counts, including 64-byte, 512-byte, 1,024-byte, and 4,096-byte frames, as well as burst-mode performance.

The results showed that FDDI adapters could provide major performance gains, but that results varied sharply by adapter, driver maturity, client protocol, frame size, and LAN environment. The lab found that the best EISA adapters delivered their strongest performance with larger frames and multiple clients, conditions that better reflected server-backbone use than small-packet desktop traffic. Performance differences were especially important because the adapter, not the FDDI ring itself, often became the limiting factor in server throughput.

Madge Networks’ Smart 100 EISA Ringnode/FDDI adapter received a Data Communications “Tester’s Choice” designation in the EISA category. The article highlighted Madge’s ability to handle both Ethernet and token-ring traffic effectively, with particularly strong results in the weighted performance analysis. Cisco’s EISA adapter also performed well and was competitive across several tests, while IBM, Network Peripherals, SysKonnect, and 3Com each showed useful capabilities but varied in throughput, driver support, or configuration options.

In the SBus category, Cisco Systems’ FDDI SBus Adapter-Turbo received a Data Communications “Tester’s Choice” designation. The lab found that Cisco’s SBus adapter delivered strong performance for Sun workstation environments and generally outpaced Network Peripherals’ SBus adapter in the test scenarios. Network Peripherals remained competitive in some UDP and TCP measurements, but Cisco’s adapter had the stronger overall showing.

The article also emphasized that FDDI was not simply a matter of installing faster cards. Driver quality, frame-size handling, operating-system support, protocol stack efficiency, and server configuration all had a significant impact on realized throughput. The lab noted that FDDI could deliver substantial benefits for overloaded servers and high-traffic backbones, but only when the adapter, server, and client environment were tuned to take advantage of the available bandwidth.

Overall, the test concluded that FDDI adapters were a practical cure for serious bandwidth constraints in high-demand LAN environments. For organizations struggling with congested Ethernet or token-ring backbones, FDDI offered a major performance upgrade, especially for shared servers supporting many clients. However, the differences among products were significant enough that adapter selection, driver maturity, and real-world workload testing remained essential.

Solutions Tested

EISA

Cisco Systems, Workgroup Business Unit — CDDI/FDDI EISA-PC Adapter
IBM — FDDI Fiber Base EISA Adapter
Madge Networks — Smart 100 EISA Ringnode/FDDI
Network Peripherals — FDDI EISA Fiber
SysKonnect — SK-Net FDDI EISA Series
3Com — FDDILink-UTP and FDDILink Media Module-F

SBus

Cisco Systems — FDDI SBus Adapter-Turbo
Network Peripherals — NP-SBS-S10