The Great ECN Brain Dump

guide for new student employees


The ECN Network Map

So, you're the lucky person who gets to work on the ECN Network Map! What's that you say? "What's the ECN Network Map?" Yeah, that's what I figured. Marian assigned it to you without much explanation of what it is or how it works. I don't know how much you know, so I'll give you a little background.

The ECN Network is a bi-annual project to document the approximate number and location of every computer on the Engineering Computer Network. The past network maps are presently hanging in the hallway outside MSEE 104, but I don't know if they will still be there by the time you read this. Before you do anything else, take some time to examine all of the old maps. Notice the changes from map to map, and pay attention to how the network has developed since its beginning.

The basic method for representing the network is to use icons for each major classification of computer. You can see from the keys on the more recent maps that there are icons for X86 machines (Windows and Linux), Macs, Suns, and assorted other workstations and servers. The actual network is broken into subnets, which are groups of computers, usually located geographically near each other, that share similar IP addresses. For example, the computers with the addresses "128.46.154.175" and "128.46.154.110" are both on the "154" subnet. On the map, subnets are represented by dotted lines with computers attached to them. Your goal when designing the map is to display every subnet and the machines on each subnet. The subnets should be grouped geographically as much as possible, meaning that the MSEE building subnets should be together, the ME subnets should be together, etc. You also need to show how the subnets are linked to each other and to the central network, all with as few overlapping lines as possible.

The Bad News

Putting together the Network Map can be a real pain. You have to painstakingly scan through a printout of every computer on the network, keeping track of how many machines of each type are on each subnet. Then you manually update each subnet on the map using AutoCad. Often this will require drastic reorganization of the map. Some machines are shared by multiple subnets, and you'll have to keep track of those. Additionally, you have to find out which "special" machines on the network warrant their own icons with their names listed on the map. Sometimes the distinction is arbitrary, so you'll have to ask around. Next, you'll want to double check your numbers against an alternate listing of machines, and you'll learn that the numbers are inconsistent. You won't be able to reconcile the data, and it's up to you to figure out why and make a decision about which numbers to use. You'll also want to send updated copies of the various map sections (ME, ECE, Potter, etc.) to their respective ECN site specialists and ask them to confirm the numbers. Remember, almost nobody cares about this project except for you, so if they bother to respond at all, they'll usually miss important mistakes. Once you've updated the numbers to your satisfaction, there's more bad news: in the time it's taken you to get this far, the numbers will have all changed because the network is dynamic.

The Good News

Despite all the bad parts about making the map, there are a few things that will make it slightly less painful. First, although the numbers are constantly changing, the overall map changes less each year than it used to. That means that you can start with the previous map and gradually add changes instead of starting from scratch. Second, Mark Senn wrote a script that will generate a list of every subnet and a count of the various machine types on each subnet. It still suffers from inaccuracy problems (resulting from the original list, not from his script), but it's a lot better than counting every machine by hand. Third, Josh Harley, the site specialist for Civil Engineering, is on your side. He was one of the original creators of the Network Maps, and he can help you a lot. Finally, (and this is the best news) complete accuracy is not expected. There is no way to get an up-to-the-second count of every machine on the network. The goal is to create a general snapshot of what the network looked like at the end of a two-year period. If you make a fair representation, everyone should be satisfied.

For more detailed instructions about creating the network map, continue to the next page.