Perhaps your school system has waited until now to start planning for a
districtwide computer network infrastructure. And perhaps the task of building
it on a tight budget has fallen on your shoulders. Well, don't panic. As
a latecomer, you have the benefit of learning from others who have been
through this process.
My own experience was not atypical. I was hired as the sole technology
employee for Saugus Union School District in Santa Clarita, Calif., in March
1996. At that time, the district was using hardly any network technology
at all. The district office had an Apple Macintosh server running basic
e-mail and backup services -- yet it only backed up Macs, so the Windows-based
PCs in the business and payroll departments were out of luck.
Without a proper network, employees had come to rely on so-called sneakernet
-- walking to a colleague's office and handing the person a floppy disk
-- as the sole means of exchanging files. The lack of a network also led
to rampant duplication of work and data. For example, the personnel and
payroll departments maintained no less than seven independent databases
and spreadsheets, all containing similar information and with a great deal
of data duplicated between them. Efforts to keep all this data synchronized
were both inefficient and ineffective.
The situation was similar at each of the 11 school sites. Typically,
office staff shared two Macs connected to a LaserWriter printer via slow
AppleTalk cabling. The remaining personal computers in the schools -- including
those in classrooms -- were not networked in any way. Some schools had chosen
to standardize on Macs or PCs, while others opted for variety. Often, this
decision was influenced by the personal tastes of the parents and staff
members who were actively supporting the school's technology program.
Obviously, our work was cut out for us: We needed to build a districtwide
computer network to give Macs and PCs throughout the school system access
to e-mail and scheduling, the Internet, file sharing, network printing,
and district databases.
To make this project manageable, we decided to divide the implementation
into two phases over a two-year period, with a total budget of around $500,000.
In phase one of the project, we would install the key central-office networking
systems, which would later support the entire infrastructure. During phase
two, we would wire and install the individual local-area networks (LANs)
at the school sites and then link the LANs to form a wide-area network (WAN)
with a hub-and-spoke architecture.
Starting small
The first phase of the project ran from March to December 1996 and cost
us about $100,000. In designing the network architecture, we kept four key
goals in mind: manageability, reliability, performance, and scalability.
Manageability was the single most important issue to us. With only one
technician on hand, we needed to be able to manage all key services of the
system from anywhere in the network. Second, the systems and servers had
to be reliable and stable, because we did not have the manpower to chase
bugs and reboot crashed servers all over the school district. Third, our
relatively small budget meant that we would have to squeeze maximum performance
and utility out of the available hardware, using each server to provide
several services at once. Finally, the system would have to be flexible
and scalable in order to accommodate the network's inevitable growth.
After evaluating our options, we decided to base our network services
on Novell NetWare. NetWare was a good match for an enterprise of our size
in terms of manageability, performance, and reliability. It also scaled
well, which would be well-suited to our multi-server, multi-site needs in
the second phase of the project.
Having installed an initial NetWare network at the district office, our
first task was to choose new e-mail and scheduling software. Our existing
e-mail system was adequate for basic internal messaging, but it lacked the
ability to exchange Internet e-mail, and it offered no scheduling capability.
An integrated e-mail and scheduling package made the most sense, and after
some brief experimentation, we decided to go with Novell GroupWise. Once
we installed dial-up e-mail access for the office managers and child care
directors at the school sites, the district communications architecture
began to take shape.
Soon, with the addition of a 64 Kbps connection to the Internet, the
district office was online. This paved the way for establishing a web site
for the school system, which accomplished two initial goals: First, to show
the community that we were in the thick of the technological revolution;
and second, to impress the school board, whose continued support was crucial
to the successful completion of the project. The old Apple server proved
adequate for this task and -- more important -- didn't cost a dime. After
several months, as speed and compatibility with our other systems became
more important, we switched to Novell's web server.
The last step in the first phase of the project was to implement basic
dial-up Internet access for the school sites. We added this interim capability
to please the board and to rally the teachers for the big push to districtwide
networking, which was soon to come. The implementation was simple: Each
school received one computer with a modem that teachers could use to log
on to the Internet via the district office network. This solution, which
made use of a Shiva LANRover remote access server, also enabled traveling
administrators and homebound employees to access our network and databases
while away from the office.
Thinking big
By January 1997, our district was still behind many other school systems
in terms of technology implementation, yet we'd made incredible progress
in less than a year. We had gone from a technology-starved district to a
collaborative computing environment with a strong Internet presence in just
10 months, and now we were ready to push these services out to individual
school buildings
This second phase of the project ran from January to October 1997 at
a cost of $390,000. As part of the proposal, my department received an assistant,
bringing the district's total technology staff to two.
Wiring the schools was the first step in the implementation of the wide-area
network. The scale of this undertaking was such that we contracted out this
part of the project to Lucent Technologies. In just a few months, Lucent
wired more than 360 rooms with more than 450 drops.
We decided to install and configure all the network equipment ourselves,
however, both to save costs and to build in-house expertise. Our design
called for placing a router and a network server at each school site. By
giving each school its own local server for functions such as e-mail and
file storage, we were able to conserve bandwidth across the wide-area network
and avoid a single point of failure. We used Compaq Prosignia servers configured
with 128 MB of RAM and dual 4.3 GB ultra-wide SCSI hard drives for speed
and reliability, running GroupWise, ManageWise, and NetWare Web Server.
Each server had its own 8 GB tape drive for regular back-ups.
We installed PC and Mac workstations in the schools on an ongoing basis,
as soon as the wiring in each school was completed. In the six months following
the start of the project, we set up workstations for more than 370 users
throughout our 11 schools.
Connecting the 11 school sites to the district office -- and thus to
the Internet -- could potentially have been a very costly undertaking, with
monthly recurring costs for 11 dedicated data lines. Fortunately, we were
able to take advantage of a Pacific Bell program called Education First,
which provided each of our schools with a full-time 128 Kbps ISDN connection
to the district office. Under this program, installation fees and access
charges were waived for the first year, after which we paid a flat $45 access
charge per line each month -- with no usage limits. We would have preferred
to avoid recurring costs altogether by using wireless links for the wide-area
network; however, this solution would have required line-of-sight between
the radios, which was not possible in the canyon where our school district
is located.
Solving problems
Internet access was the buzzword that was most important to the parents
and the school board, and it was probably the primary reason they wanted
the network installed. But there were several issues that had to be resolved
before the Internet could work for us. First was the issue of filtering;
we needed to control access to sexually explicit, violent, or otherwise
destructive material on the Internet. Second, we needed to conserve the
network's limited bandwidth, especially between the schools and the district
office. Third, the small number of IP (Internet Protocol) addresses allocated
to our district meant that we had more workstations than addresses that
could be assigned to them.
For filtering, we chose SurfWatch proxy server from Spyglass. This software
runs on top of Microsoft Proxy Server and Windows NT Server 4.0 and acts
as a gateway that attempts to keep unwanted Internet content from entering
our network. To keep the kids from simply disabling the proxy server settings
and getting past the filter, we set up an IntranetWare server at the network's
border that acts as a firewall and filter, using the built-in NetWare Multi-Protocol
Router. No Internet traffic to or from our network is allowed to pass through
the firewall unless explicitly permitted by the filter.
Around this time, we also upgraded our district's Internet connection
from 64 Kbps to 384 Kbps. Yet a lack of bandwidth was still a concern for
us, especially when many kids in a lab environment accessed the Internet
simultaneously, which could often swamp a school's ISDN line. To solve this
problem, we installed Novell's BorderManager at each site. These local proxy
servers gave a much-needed boost to each school's perceived Internet access
speed by saving frequently accessed web pages in a cache at the local server.
Using a proxy server at each school site also gave us a virtually unlimited
number of available IP addresses, thus taking care of our third concern,
as well.
Working the web
One of our goals for the network was to use web services to increase
the availability of information throughout the district. Our ambitious plans
called for a web server for every school -- and a web page for every teacher.
The idea was to provide valuable information resources to all users, regardless
of computing platform.
To make this happen, each school has a NetWare Web Server that offers
a basic home page with a picture of the school and a listing of teachers
who have published their web pages. Those teachers who know HTML can simply
upload their files into their home directory on the server. For those teachers
who don't know or don't want to learn HTML, we've provided an Internet-based
web site builder program. It works like this: The teachers simply access
our home page, locate the builder program, and fill in a form. Their home
pages are then generated automatically and placed on their school's site.
To top it off, we add links to their pages from their school's home page.
We also provide many other web-based services from the district's central
web server, including forms that can be filled out online and printed out,
online technical help for users who need help configuring their computers,
and documentation for various programs -- all searchable. We offer all these
tools to help reduce the support burden on our shoulders and empower our
users to become more self-sufficient. Also, each school web site has a link
to web-based e-mail access, so that employees can retrieve their e-mail
from anywhere on the Internet using just a browser.
Recently, we've been exploring online applications. We now have an online
work order system on which users submit work order requests. The maintenance
department reviews them and approves or denies them -- all over the Internet.
We are also running a web-based scheduler program for the Santa Clarita
Valley Consortium, which is a group of local school districts that are providing
free training classes to teachers, parents, and students over the summer.
Users can browse the available classes and sign up online; the system checks
their schedules for conflicts, processes changes, and generates a class
list quickly and easily.
Final thoughts
While the above technologies may seem relatively unremarkable, we are
proud of putting everything in place in a short time and on an extremely
small budget. Everything was completed in less than two years on a budget
of around $500,000. For that money, we have established a flexible system
that offers everyone throughout our organization access to powerful communication
tools and services, all of which can be managed easily by one person from
any location in the network.
How did we do it? By learning the technologies in-house and leveraging
that knowledge to configure and install systems without the use of expensive
outside contractors. In my view, our school district has shown how effective
a well-trained, well-supported, and highly motivated technical staff can
be. More important, I think we've proven that a major project such as this
doesn't have to break seven figures.
We're already thinking about future projects. At the district office,
we will be evaluating Callware's voicemail package for GroupWise. If we
are satisfied, we will most likely place it in all the schools, offering
parents without computers a convenient way to contact teachers and check
assignments. In doing so, we hope to encourage these parents to stay involved
with their children's education. Another upcoming project is to consolidate
the individual student databases at each school into a central, client-server
offering. We also have a new school opening in June, which will require
the same tools available elsewhere in the district. In short, we have our
work cut out for us. And if we can do it, so can you.
James Klein is director of information services at Saugus
Union School District in Santa Clarita, Calif. | 
