Networking

Network planning: Pay now or pay later

You can't exactly ask others how they've configured their networks (etiquette and all). But you're curious about how well you've done with your own. Here's how Hamilton College approached its systems design.


For almost 15 months, I’ve been involved in a network electronics upgrade at Hamilton College, a small liberal arts college in upstate New York, where I manage the network services department. I learned a lot from this project, including the importance of proper planning.

The importance of proper planning
The first step in planning any project, IT-related or not, is to determine its need. If something is working fine and will continue to provide the required services until the next millennium, there is probably no need to make changes. However, when a system begins to show signs of age, or sag under heavy loads, it’s time to decide what to do.

At Hamilton, we had a network that was originally installed in 1994 and 1995. It consisted of two Fore Systems/Alantec Powerhub 7000s with 10 Mbps fiber connections to stacks of 3Com hubs in all 58 campus buildings. These stacks of hubs ranged anywhere from 24 to 96 ports sharing the 10 Mb uplink. The campus servers were connected via a Fast Ethernet connection to a 10/100 switch. The two Powerhubs were connected to each other via an FDDI ring running at 100 Mbps.



I want my MP3s!
In 1995, this was state-of-the-art, fast technology. In 1998, it began to show its age, and it began crashing on a fairly regular basis due to heavy loads. Outages prompted the need for an infrastructure upgrade. After all, a college network manager can handle only so many irate students not being able to download their favorite MP3 files!

After identifying the need for network modifications, the next step in the planning process was deciding exactly what could be done to improve the system.

A variety of ideas, such as simply installing more 100BaseFX modules in the Powerhubs, was discussed. Ultimately, we decided on a more drastic course of action for a number of reasons. We chose to completely replace both Powerhubs and install new equipment at the edge of the network to increase each building’s bandwidth.

Replacement versus upgrade
Why replace the network rather than upgrade it?

Mainly, our equipment was old and upgrading it was an issue. At the time, the maximum possible connection to a Powerhub 7000, according to the sales folks, was a 155 Mb ATM connection or a 100 Mb Ethernet connection. We knew that we wanted to avoid ATM and its complexity. We also didn’t want to be stuck with 100 Mb Ethernet and face another major upgrade in only a year or two.

Select a vendor
At this point, we began the next step in the process—vendor selection. We began speaking with the usual suspects—Cisco, Nortel/Bay Networks, Cabletron, and 3Com. In the end, we chose 3Com for a couple of reasons. The first reason was 3Com could provide a clean and reliable solution at both the network core and at the edge. The second reason was we already had a large amount of 3Com network equipment on campus, and we were happy with the reliability.

We ended up choosing one 3Com CoreBuilder 9000 to replace both Powerhubs. At the edge, we decided not to completely replace all of the equipment due to cost restrictions. We instead planned to install a 3Com Superstack II Switch 1100 (10 Mbps switch with two 100 Mbps uplinks), each having a 100BaseFX module that would be connected to a 100BaseFX port on the CoreBuilder 9000.

For our server connection, we planned for two Gigabit Ethernet modules in the CoreBuilder that would be trunked to a 36-port Superstack II Switch 3900 (10/100 switch).



This equipment configuration provided us with a number of benefits. For one thing, it is very easy to upgrade a connection. If we need more bandwidth to a building, we can replace the module in the Switch 1100 with a Dual 100BaseFX or a Gigabit Ethernet module. The CoreBuilder itself has the capacity for new switching fabrics to allow for greater capacity at the core. The server connection is upgradeable to 3 Gbps, if the need ever arises.

These upgrades can be very easily accomplished and can be completed without purchasing much additional equipment. To make the core very reliable, we have redundant switching fabrics and redundant Enterprise Management Engines, which are the brains of the CoreBuilder.

Five months: time well spent
This process of determining the need for an upgrade, choosing a vendor, and selecting equipment took almost five months. But in those five months, we took the time to properly plan the network for both today and the foreseeable future. We know that it will require continual care and maintenance, but we are confident that we will be able to use the existing equipment for quite some time and that it will be able to grow with our needs.

This is where the importance of proper planning comes in. If we had decided to take the easy route and to just upgrade our old Powerhubs, we might have put both time and money into something that would be useless to us fairly soon. By taking the time to carefully plan and investigate our options, we have made an investment that will serve us well.

Don’t change that channel!
In the next part of this series, I will discuss the actual implementation planning, as well as some of the problems that we experienced and how we worked around them. In the third part, I’ll explain how we maintain the network infrastructure, what we plan for the future to ensure the reliability of the infrastructure, and how we plan to improve the network down the road.

Scott Lowe is an associate director and network services team leader for New York’s Hamilton College. He’s earned MCP+I and MCSE certifications from Microsoft.

If you'd like to share your opinion, please post a comment below or send the editor an e-mail .

Editor's Picks

Free Newsletters, In your Inbox