How we did it
|
|
|||
|
|
|
|
|||
|
|
We focused on the ability of these products to help find and resolve network infrastructure problems, including cable faults. We judged each product’s accuracy, range of tests, reporting capabilities, ease of use and documentation. Just as you’d expect, price was also a consideration.
The test bed network consisted of six Fast Ethernet subnet domains connected by Cisco routers, back-to-back Visual Networks DSU/CSUs and a Covad symmetric DSL Internet link. The network also incorporated Lucent’s WaveLAN, Proxim’s RangeLAN802 and RadioLAN’s 10BaseRadio to provide wireless links. The range of servers and clients included Windows 98/ME/NT 4.0/2000, Red Hat Linux 6.2, Macintosh System 8, Solaris 8.0 and OS/2 Warp 4.0. The network’s transport layer protocols were TCP/IP, IPX/SPX, NetBEUI, AppleTalk and SNA.
To test their ability to find cable faults, we deliberately connected these devices to miswired cables, broken cables, too-long cables (more than 300 meters for 100Base-TX) and insulation-stripped cables. We also exposed the devices to networks with levels of utilization from zero to 90%, jabbering, broken network adapters (we label and save them for just these sorts of tests), misconfigured full/half duplex links, unresponsive Dynamic Host Configuration Protocol and DNS servers, short Ethernet frames and a variety of network and server connectivity error situations.
