Over the past year and a half we have helped a few schools in the vicinity of Murray Hill, New Jersey connect to the Internet using non- trivial bandwidth. What was originally a volunteer effort has evolved into an interesting program that continues to generate interesting problems. Along the way we found ourselves dealing with Internet traffic congestion, local networks in schools, public domain software, training, microwave links, several types of servers, school boards, politics, communities, and - oh yes - the sudden surge of awareness of this thing called the Internet.
Several years ago a small number of researchers based in Murray Hill were provided with FDDI links to their homes using RdarkS fiber rented from the local cable company. While I wonUt detail what it is like to have a connection like this, I will point out that I was unsuccessful in lobbying my wife to make a move to the region and still connect at telephone rates.
The standard pattern of unusual social connections occurred and the Superintendent of the Berkeley Heights Middle School became aware of the fiber program and asked if it would be possible to connect his school. Pat Parseghian and Norm Schryer of the Labs undertook the effort to link the school to the Internet at Murray Hill, while a local computer store worked on the internal school network and the local cable company (then Suburban Cable) supplied the fiber link.
A surprising number of things went right and link came up in August of 1994 with high speed links were demonstrated over a fast backwater of the Internet known as XUNET with the actual connection to the Internet coming from us in Murray Hill. Around that time several other workers including Gerry Ramage, Wenling Hsu, Minsky Luo and myself volunteered help.
The middle school had a computer room with about twenty four PCs (486sx/33MHz) each with 4 Megabytes of RAM living on a Novell network. The software, hardware and network have each generated bits of grief. A bottom line learning is that consulting with a school before they submit their technology plan is very important.
The performance of the Novell network was terrible and we were forced to abandon it using a loaned Silicon Graphics Indigo as a local router (note that a FDDI speed router is very expensive). Suddenly we jumped from sub POTS (plain old telephone service) to one megabit per second data rates assuming we were connected to an unloaded server.
The next problem to deal with was (and is) bad public domain software, a variety of hardware and software configurations, and a lack of RAM on the machines. We were forced to use Netscape 0.94b for nearly a year as it was the only version of Netscape that would work with such limited memory with the machines as configured at the school. Eventually the machines were upgraded to 8MB, but that is still insufficient for web-related work these days. To this day we still have problems with machines hanging and crashing.
Any of you who regularly surf will note that data transfer rates have a strong time of day dependence. Many of the worst times, of course, are during school hours. Compounding the problem was the fact that the school was sharing a T1 link with workers at our large installation in Murray Hill. A solution to this problem is still under development, although a partial solution has been provided in the form of a caching proxy server with a large amount of disk space. The idea is that a teacher can pre-surf the sites that he or she wants the students to visit. Access to anything that lives on the cache will be very quick. Of course there is a need for cache control tools and we still have reliability and performance issues.
Student and teacher training has become a major issue. We have learned that hands on training sessions are essential. Teachers tend to divide into three groups - (1) those who are already using the Internet, (2) those who have to be shown that it is important to them and (3) those who seem to be against anything that involves a computer. It is important to identify teachers in the first group and use them as evangelists. Helping them get personal connections to the Internet at home has proven to be extremely useful.
Pornography has also been a major issue. Some of the schools weUve approached view it as a show stopper and refuse to allow students access, other schools feel that honor guidelines are sufficient and that the potential payback exceeds the risk. Some internal work has produced a rating system that only allows positively rated content, but some of the administrators demand something that is 100% pornoproof. Of course courses can be designed to minimize the amount of random surfing and X-rated material that isn't on the local cache will arrive slowly.
A year into the project these and administrative problems insured the Internet connection was only being used by a limited number of teachers and students (although a few teachers were using it to help prepare lesson plans for other classes). Another school was banging at our door, so being fools for punishment we agreed to put in a similar link.
The New Providence Middle/High School has a few dozen Macs and Windows machines on ethernet LANs. One of us was interested in trying the Macs out due to the problems we had experienced with the Windows machines (the Rit canUt be worse than this theoryS). Total time to configure two dozen Macs and have them solidly talking to the Internet was on the order of a few hours rather than a few weeks.
Although the Macs were very easy to configure and easy to add multimedia applications (sound was difficult at the other school due to non existent and/or outdated sound cards), there was a performance hit caused by MacTCP, which effectively limited the machines to 200 kbps performance compared to PCs, which regularly saw nearly megabit performance. This has recently been corrected with the introduction of a new Apple provided TCP stack called Open Transport 1.1.
New Providence is already running several Internet based classes during school hours and opens up to the local community a few nights a week. The Berkeley Heights school is now expanding their internal offerings. Both schools have volunteered to be model schools and regularly host visits by interested third parties.
Of course the Web is not the Internet and real two way communications (synchronous and asynchronous) are important. Email and netnews are the most obvious. We have not provided a netnews feed, but have installed SMTP/POP3 email servers locally at the schools. Email use by students has not been operationally sorted out, but a subset of teachers have access. Several other communications projects are now underway.
While all of this was going on, we were interacting with several schools that we could not help with access. We installed a platform that would allow schools to have server space on a well connected/maintained server as a public service. The idea was that we could reach communities where some fraction of the population could connection via an online or Internet service provider. We were coming to recognize that the effort to get a reasonable connection to the Internet to a school or community was a major hurdle to many communities and the thought of lowering a barrier to help jump start a grass-roots effort was attractive. We also werenUt interested in becoming an ISP!
Sixty geographically distributed teachers, with a minimum of an email connection, were approached (fifty of which had participated in an AT&T sponsored program for outstanding teachers) and offered all of the server space they wanted with expansion to local newsgroups and chats in the future. Eleven teachers indicated an interest and a mailing list was started to establish a communications link.
We quickly learned that teachers are very busy and finding time for projects like this without a greater level of support was too much to ask. Four actually initiated projects in their schools (two working on non-trivial content). At this writing one of the sites is being held up by the local school board who canUt figure out if this is good or bad, another teacher found a local ISP who was willing to do hands on training in the classrooms and host the resultant pages for nothing and the remaining projects seem to have died.
The bottom line is that an enormous amount of training and support is needed for such efforts. While we still feel it is important to reduce barriers by offering server space to schools who currently are not in a position to run their own server, we recognize that a strong grassroots effort is essential. Someone has to devote quite a bit of time and expertise. Students can do amazing things once they have been given tools and techniques, but this usually doesnUt happen by itself and some adults will even place barriers in the way of the kids.
An aerial packet assault
Our latest project hasn't borne fruit yet, but is interesting. By last Summer we were convinced that a community based server made sense as it provided a mechanism to couple local talent and effort to the access and hosting problem. We wanted to see if our barrier removal technique would couple to a community project.
The problem was finding a viable community with enthusiastic established organizations. We finally began to work with the schools in Plainfield, New Jersey. Plainfield is interesting in that, while it has some serious problems, it is clearly a community (many communities in New Jersey are really commuter communities where there is little coupling among the residents).
We have installed a 2.4GHz 2 megabits/s microwave link between our location and a school about five air miles distant. While this may seem to be an unusual solution, consider that the microwave equipment is a one time purchase and can be much cheaper than leasing a T1 connection over time (it can even be cheaper than 56k leased lines or ISDN in some situations).
We are hoping to establish microwave links from our target school to other local schools, the library,and possibly some public access locations. If one is blessed with line of sight communications (we are in this case), the incremental cost of linking other nearby locations to a centralized point is low. We happen to be providing a T1 equivalent access link to the first school, but this could be any reasonable link to the Internet.
While weUre just helping out with an Internet feed and some training (weUll help them with their Macintosh network and provide some teacher training), we have established a link with an active community organization known as RFrontiers. We are confident that these folks will get local businesses, government, citizens, etc. on board. Talks are proceeding with a few ISPs to bring inexpensive POTS service for members of the community and a school based service organization may be refurbishing discarded PCs and Macs for resale to the community. If a local Internet based netnews based BBS springs up, a low end PC or Mac Plus running a VT100 emulator could give many people community access at very modest prices.
So what have we learned?
1. Bandwidth is important to schools if they are using the Web and want to keep the attention of the students.
2. Bandwidth is still expensive at this point and simply providing something like a T1 to a school is not the whole solution. There are several interesting solutions on the horizon.
3. Training is a major issue. None of this is simple yet.
4. Local hardware, software and LAN configurations are major issues. Some machines are much easier to work with at this point.
5. Local communities want and need control and should set the standards (particularly true in the case of pornography).
6. Tools are rapidly improving, but there are still barriers to publication even if free server space is offered. Some of these barriers are institutional.
7. Community involvement may be the only way to create a critical grass-roots interest.
8. Doing this can hose all of your free time! Burnout can be a major problem, so it is important to have a critical mass of volunteers.
9. Communities need to have knowledgeable volunteers advising them on their technology plans. We have come up against a few good plans, but the majority seem to be clueless.
10. Top down efforts will probably fail for a variety of reasons, while grass roots efforts have their own set of problems (sustainability, money, expertise, etc.). Some combination of the two may work.