One Solution to Setting Up a Whole-School Network ...

The Starting Point:
1. We recently had our school networked with fibre optics, switches, hubs, etc.
2. In the next three years we are expecting to have well over 200 computers on the network.
3. We currently use six phone lines and six Internet accounts to provide Web access around the school. (Even though we now have a 'centralised' network structure we believe it is best to stay with this 'multi-account' arrangement for the short term.)
4. We use Wingate Version 2.1d as our Internet router.
5. All servers will be based on NT4.
6. Our school layout consists of six large 'blocks' of classrooms and an Administration 'block'.
7. Our school will change to a "Middle School" in 2001 with a reduced student population.

The Design:
Our first intention was to set up a 'multi-domain' structure for each of the 'blocks' with a central server controlling access and providing a degree of isolation between each domain. We very quickly found this sort of design to be outside our capabilities and opted for a single domain structure. (I imagine this will change in the future as demand for a more 'secure' system develops.)

Sub-net Mask
Our first hurdle was the "255" limitation on the number of computers we can have on a network with a sub-net mask of "255.255.255.0" After several hours grappling with the concept it finally fell into place. Hopefully the explanation below will help clarify the situation  ...

Network addressing is made up of four 'octets' (8-bit binary numbers) represented as "abc.def.ghi.jkm" (where "a" to "m" are decimal numerals). The "sub-net mask" is used to determine whether sections of the IP address are used for "Network ID", or "Host ID". If the binary form of the mask has a "1" then that place in the IP Address represents the network ID. If the place is a "0" then that place in the IP Address represents a Host ID. (The situation is a bit complicated by the fact that the numbers are usually written in decimal form.)

Example1

Example 2

Note: It is possible to manipulate the sub-net mask so that only part of an octet is used for this sort of Network ID/Host ID delineation, but it gets quite difficult to work out.

DHCP IP Addressing:
When setting up individual IP addresses the choice is between "dynamic" addressing and "fixed" addressing. The arguments for each are quite convincing. We chose to go with "dynamic" addressing to simplify the situation when machines are added to the network over the next few years.

Wingate:
In the past we have found that Wingate2 and Wingate3 do not co-exist well on the same network. We had further problems when setting up the system because someone had installed Wingate3 without our knowledge. It appears to interfere with the DHCP server and prevented clients from obtaining an dynamic IP address. (Half a day later, after isolating the Wingate3 system, the network was  working well.)

The new system consists of six Wingate 'routers' spread throughout the school running on Window95 and NT4. Each is connected to a single phone line and operates on a single Internet account. Each Wingate server has a different 'fixed' IP address. The local browsers are set up to 'point to' each Wingate server using its IP address (not the name "Wingate", which would be the usual practice).

Suitable References:
I recommend that anyone responsible for managing any NT4 system purchase the book - Using Windows NT Server 4 by Drew Heywood. Published by Que. ISBN: 0-7897-1612-7

For comments and queries contact me at:

Return to Home Page