How to Tame Big Networks

Whether you're designing, simulating, or maintaining extra-large networks, smart software tools can help.

Edmund X. DeJesus

Quantity has a quality all its own. While running a 10-node network might be challenging, and handling a 100-node network might be absorbing, designing and maintaining a 1000-plus-node network are harrowing tasks. The larger the network, the more likely it is to contain a variety of hardware platforms, OSes, connection schemes, and protocols that span countries and continents. In short, bringing order to this potential chaos is more than a single, mortal administrator can handle.

The mega-network administrator's primary concern, of course, is how to foist off this crushing burden. The goo d news here is that tools exist just for t his purpose.

For example, designing a large network or adding onto your existing behemoth can be taxing, since you have to consider so many parameters, including throughput, location, compatibility, size, and budget. New graphical tools let you assemble big networks from databases of components as if you're piecing together Tinkertoys -- and smart Tinkertoys at that, since some of these tools can tell you if you're making blunders such as using an underpowered server for an engineering workgroup.

If instead of designing a network from scratch you must integrate two existing networks, modeling tools can give you the big communications picture. Some products make you get your hands dirty and fill in all the details about the components yourself, while others, using "automatic discovery," quiz the myriad components and construct a model for you.

With so many interconnected components to watch over, maintenance might be your biggest headache. Most network-administration centers look l ike NASA Mission Control, with all their flickering monitors, each tuned to specific hardware components. Enterprise management systems can reduce the number of monitors you have to stare at each day, and they also consolidate information that logically belongs together. Event-correlation capabilities can even figure out automatically that, yes, it was that iffy router on the thirteenth floor that just triggered the 42 irate calls the help desk just fielded.

The harried system administrator is no longer alone in the battle to keep a company's information electrons flowing efficiently. And that can bring some quality to the business of handling all that quantity.