ImageNet's Computer Aided Network Engineering (CANE) is the same type of modelling tool as Zitel's NetArchitect: both products require an inordinate amount of manual input and user intervention.
Nonetheless, with the addition of a few key features, CANE, Version 2, will be poised to fill the gap between high-end modelling tools and isolated end-to-end tools. We took a peek at this latest 32-bit version.
To be sure, CANE holds its own in the world of "isolated" modelling tools. Along with NetArchitect, CANE requires you to recreate your network from scratch, selecting clients, hardware, and software from a library of devices. Fortunately, ImageNet plans to add an auto-discovery feature to its next major release. Until then, you'll have to create specific devices and add them manually to your network map. The product supports only a few devices, including those made by Cisco, 3Com, and Compaq.
Despite having to re-create your network, CANE is a remarkably intuitive tool. The controls are easy to understand, and the detail available on every device is unsurpassed by the products in this comparison. CANE offers specifics such as the number of slots in a chassis, the software version, and the maximum number of ports per card. Of course, you have to fill in all of this information by hand.
The overall design of CANE allows for a clear representation of your enterprise. You can create network maps from the top down and/or bottom up, and you can represent a big picture of the network, including countries, states, and cities. From there, you can drill down into specific buildings and floors within those areas. Creating traffic links was easy, too. You have to know a little about the traffic you're applying, such as number of packets, packet size, and active and sleep times for the conversation. Otherwise, it's painless to create an accurate model of your conversation.
One of CANE's most impressive features is its simulation engine. The real-time bar and dial graphs show not only the simulated packet rates and media utilisation but also offer perspectives on each component in the model. These features were easier to use and more robust than even NetArchitect - which had some pretty impressive dynamic-simulation graphs. With CANE, every component in the model can be individually graphed throughout the simulation. So if you're modelling a client at a remote location connected to a server on your local network, the product will show the frames per second and utilisation of the client. It will also show the remote network, remote router, WAN link, local router, and the server itself.
As indicated by the number of helpful online resources, ImageNet is clearly not new to addressing the learning difficulties of its product. Resources include the Scott Bradner report on network device performance, the LAN Times Networking Encyclopedia, and a buyer's guide.
Pricing for CANE starts at $US4000. ImageNet can be reached at www.imagenet-cane.com.