NetRule Ably Predicts Net Growth, but Setup Is Tricky

Analytical Engines Inc.'s lightning-fast NetRule 4.0 network modeling and performance simulation tool is a good guide for planning network growth.

Analytical Engines Inc.s lightning-fast NetRule 4.0 network modeling and performance simulation tool is a good guide for planning network growth. Senior IT executives and network planners should put this product in their tool kits before spending one more dollar to add capacity to the network.

NetRule, like competitors such as Opnet Technologies Inc.s IT Guru and Compuware Corp.s EcoPredictor, uses a simulation engine to predict network performance based on factors including traffic load, protocols and the number of users, along with the kinds of transactions they conduct. NetRule runs on any platform that supports the Java 1.2 run-time environment. It is priced at $15,000 per server; the Multitier module is $5,000 more.

Our biggest problem with this update, which was released last month, was that we had to add all network elements by hand. Even though the products utilities make it easy to reuse common elements such as PCs and 100M-bps LAN segments, it would be much better if this information could be imported from an inventory system.

NetRule, with its Multitier module option, goes one better than the pack. In eWeek Labs tests, the product enabled us to add real-world complexity to a test model by showing complex processes that happen at random times as transactions are processed. For example, we ran a simulation during which users made a request to a server that, 10 percent of the time, sent a large amount of traffic to a print server.

New failure analysis features let us easily identify capacity problems, and we used new server overhead parameters to more accurately emulate the latency caused by heavy loads on various servers in the test design.

Tedious setup

Despite attempts to keep it as simple as possible, it was easy for us to get lost in the complexity of setting up simulations. Therefore, although the product runs performance tests with surprising speed—often under 10 seconds in tests, running on a modestly equipped Dell Computer Corp. 550MHz OptiPlex system—setting up the simulation isnt any easier than using other products weve tested.

This is a small price to pay, however, because loading the network details into NetRule is one of the first and best steps IT managers can take to get on the road to planning a network build-out. As was true for previous versions of the product, it was easy to set up equipment and segment profiles that were simple to add to a network document. After adding these network elements, we were able to change the performance characteristics by clicking on an object and modifying the property sheet.

IT managers should be aware that NetRule relies on modeling and sampling instead of discrete event calculation to arrive at its predictions of network performance. Discrete event simulators calculate network performance for each packet at every juncture in the network. In the past, these products were often considered the most accurate predictors. The drawback was that they also took ages to run complex simulations.

Traffic spotting

During tests using NetIQ Corp.s Chariot traffic generator and the IPWave network impairment emulator from Spirent Communications Inc. (a subsidiary of Spirent plc.), we were able to get traffic predictions that came within a very respectable 8 percent to 16 percent of what we expected to see. Even the most complex simulations that we ran took less than 20 seconds.

This enabled us to tweak and change network designs quickly in order to zero in on the best solutions to design problems.

NetRules failure analysis module also served us well during tests. NetRule went through the test design and, using the downtime probabilities that we assigned, was able to knock out each component in the design and then see how the network performed. As expected, these simulations barely took more time to run than the regular trials.

NetRule packaged the results in very helpful graphical reports that clearly pinpointed the places where backup routes were either nonexistent or too overburdened to handle additional traffic.