Network Computingwww.networkcomputing.com

We tested Castle Rock Computing's SNMPc Workgroup 5.0, Ipswitch's WhatsUp Gold 4.0 and MG-Soft's Net Inspector. All three cut to the chase quickly, producing solid network-status information for less than $1,000.

WhatsUp Gold, a workgroup workhorse, took our Editor's Choice award. It's simple to set up and use. It doesn't have a growth strategy or complex trend reporting yet, but these features are in the works. You can go from setup to status in a single afternoon. Castle Rock's SNMPc finished in the middle of the trio with a set of impressive diagnostic tools and the expansion capability necessary to grow into a much larger solution. More complex than WhatsUp, SNMPc is very much a traditional SNMP network-management tool, but it didn't always behave during our testing. MG-Soft's Net Inspector came in third. This 1.0 version exhibited flashes of brilliance, and will likely be a contender when version 2 ships later this year.

How We Tested
Our primary concern was the ability to quickly set up a management infrastructure that reported availability, and gave us diagnostic tools to dig deeper into the whys of failure.

We set up each product on its own Microsoft Windows NT 4.0 server, running on identical 300-MHz Pentiums with 256 MB of RAM. We wanted to test how the products performed on a small-to-medium-sized network, so we limited discovery to a few hundred nodes. We located local campus segments at Syracuse University, and over our frame relay network connecting our San Mateo, Calif., Madison, Wis., and Washington labs.

We wanted to see how these products fit in workgroup situations, and we rated price a significant factor along with usage and implementation. However, both Castle Rock and MG-Soft offer enterprise versions, and we considered this when scoring architecture for our report card. However, our tests were performed on the workgroup versions only.

The autodiscovery process was imperfect--as it always is--but showed that address filters and seed routers are just as important with a few hundred nodes as they are with thousands of nodes. We rediscovered the hard way that seed routers and address filters can prevent Internet discovery or cause the software to jump over adjacently connected nodes.

We wanted to gauge SNMP usage and error statistics for Layers 2 and 3, such as In and Out Octets, as well as discover what each product could tell us about services running on well-known and user-defined TCP/IP ports. We set up alarms and reports to indicate network health, node and service availability. We also enabled daily status reports. Additionally, where diagnostic tools were available, we exercised them to see what they could tell us about failing devices, and to determine how direct or arcane their information and usage were.