Abstract

3Com commissioned Tolly to evaluate its Fast Ethernet solutions, with the main focus on measuring the performance benefits of two upgrade approaches: the LinkSwitch 1000 10/100 Ethernet switch and the LinkBuilder FMS 100 100Mbit/s Fast Ethernet stackable hub. The project examined how each product improved bandwidth for overloaded 10Mbit/s Ethernet LANs, both when legacy 10Mbit/s client segments accessed 100Mbit/s servers and when end stations were upgraded to 100Mbit/s network access.

The December 1995 report shows that both products delivered major gains over traditional shared 10Mbit/s Ethernet. In single-server testing with 24 clients, the LinkSwitch 1000 achieved aggregate throughput of 86.01Mbit/s using 1,516-byte frames with packet burst enabled, and still delivered 36.14Mbit/s with 512-byte frames. In the same single-server configuration, the LinkBuilder FMS 100 reached 86.31Mbit/s with 1,516-byte frames and 35.86Mbit/s with 512-byte frames. Tolly notes that with Fast Ethernet to the desktop, a single client could use nearly 80Mbit/s of bandwidth, an important advantage for bursty, bandwidth-intensive applications.

The dual-server tests revealed the main difference between the two designs. With two servers attached to separate switched 100Mbit/s ports and 24 clients evenly divided between them, the LinkSwitch 1000 delivered 156.25Mbit/s aggregate throughput using large frames with packet burst, and more than 66Mbit/s using 512-byte packets. By comparison, the LinkBuilder FMS 100 delivered 81.86Mbit/s in the same large-frame dual-server test and 46.68Mbit/s with 512-byte packets. Tolly explains that this result reflects the architectural difference between switched and shared media: the LinkSwitch 1000 can scale beyond 100Mbit/s aggregate throughput because separate switched server ports eliminate the single-segment bottleneck, while the FMS 100 shares one 100Mbit/s segment among all users and servers.

The report concludes that both solutions significantly improve performance, but their best use cases differ. For single-server environments, shared 100Mbit/s Ethernet via the FMS 100 can outperform or closely match switched 10/100 Ethernet. For workgroups with multiple heavily used servers, however, the LinkSwitch 1000 offers clear aggregate throughput advantages. Tolly’s tests therefore present the LinkSwitch 1000 as a scalable solution for mixed 10/100 workgroups and the LinkBuilder FMS 100 as an effective way to bring Fast Ethernet bandwidth directly to desktops and servers.

Solutions Tested:

• 3Com LinkSwitch 1000 — 10/100Mbit/s Fast Ethernet switch tested for switched client/server throughput.  
• 3Com LinkBuilder FMS 100 — 100Mbit/s Fast Ethernet stackable hub tested for shared-media client/server throughput.