|Amherst College DSL Evaluation|
© 1999 by Stephen A. Judycki
By Stephen Judycki
Amherst College is an independent liberal arts college located in Amherst, Massachusetts. Most of its buildings are collocated on a private campus situated near the center of town. Being collocated on private land, these buildings were easily threaded with a fiber-optic backbone that gave each one direct access to the campus data network. Amherst has other buildings located nearby within the town, but they are separated from the campus by one or more public rights of way.
We investigated several options for providing network access from these remote buildings. Extending our own cable plant from the main campus proved to be cost prohibitive, given the distance in cable feet and the relatively few number of users that would be served. Ten of these remote buildings are dormitories that house a total of 250 students. Before the remote dormitory rooms were wired with data jacks we used dial-up modems, but this approach had a host of problems and indirect costs that most people are keenly aware of. Then, coincidental with wiring the dorm rooms, we replaced the dial-up modems with ISDN basic rate service. Each remote dormitory LAN now has a dedicated 128kb/sec ISDN 2B+D link to the campus data network, and the ISDN bridges we installed perform 8:1 data compression on previously uncompressed data. The implementation was fraught with problems though. For longer than I care to mention we struggled with our bridge vendor, bridge manufacturer, and the LEC over dropped ISDN connections. The problem was ultimately solved when the vendor replaced, at no cost to Amherst, its poorly designed bridges. Because an ISDN 2B+D circuit-switched data line is relatively slow and fairly expensive, we really didn't want to add any new ISDN lines. LEC-provided T-1 service is available. It's faster than ISDN, but it's also more expensive. LEC-provided ADSL service is still going through its trials and isn't yet available. So we started to look at the many ADSL and SDSL products that were entering the market.
It wasn't until PairGain introduced its Campus-HRS Remote that we saw a flexible and affordable SDSL/HDSL product that would make a DSL evaluation worth the effort. Campus-HRS works on either one or two pairs of copper wire, at greater speeds and over greater distances than other products we had looked at to date. In December 1998, one of our remote buildings - the Observatory - was undergoing renovations to prepare for the tenancy of a small academic department. The project included the installation of Category 5 data jacks, a patch panel, and a 10BaseT hub. The tenant's personal computers, upon being connected to the data jacks, would become nodes on the Observatory LAN. The linking of this remote LAN to the campus data network, using HDSL as the underlying delivery system, would be the focus of our evaluation. The LEC would still have a critical role though. We would need them to supply a copper wire path between the HDSL modems located in the Observatory and a building served by the campus data network. For this building we chose Converse Hall, because of its close proximity to the LEC's central office.
The LEC in Amherst, Massachusetts is Bell Atlantic – formerly NYNEX. We asked our Bell Atlantic account team to order a dry copper circuit for us and their response was something to the effect of “Sure, what kind would you like?" Having never furnished a dry copper circuit for this purpose they didn't know what we needed, and the terms I used to describe what I needed were not at all useful to them. They provided copies of tariff pages that described the availability of Series 1000 channels that they said included the dry copper circuits used by alarm monitoring companies. The tariff described Type 1001 as a circuit that connected two points via one central office, and Type 1002 as a circuit that connected two points via two or more central offices. Bell Atlantic explained that these Type 100x circuits have many sub-categories of circuits that aren't even identified in the tariff, and each sub-category has its own USOC (Universal Service Order Code).
We knew the distance vs. performance characteristics of the Campus-HRS Remote as well as the data rate that was needed to make DSL a viable option for the Observatory. The next step was to request a Loop Makeup (LMU) from Bell Atlantic for a Type 100x circuit between Converse Hall and the Observatory. An LMU provides the length and gauge of copper wire for all of the segments, both underground and aerial, between the two points named in the request. The LMU results indicated a total distance between Converse Hall and the Observatory of over three miles, which did not bode well for the 1mb/sec data rate we were hoping to achieve. But we tried it anyway - we ordered a pair of Campus HRS-Remote modems and had a dry copper circuit installed. Because the modem purchase was conditional upon a successful evaluation, the risks were limited to the dry copper circuit's installation cost and recurring charges for a couple of months. At this point in time we did not know about the different USOCs that were available from Bell Atlantic and did not specify one.
When the modems arrived we configured them for one-pair operation, selected a common data rate, and installed a data patch cable between their line ports. Establishing a link in this back-to-back configuration confirmed that the modems were configured properly and were operational, before introducing other variables. Next we cabled the modems to the two supplied jacks, and installed a short length of one-pair phone wire between the jacks to simulate the dry copper circuit. The jacks are wired differently, depending upon one-pair or two-pair operation, and establishing a link in this back-to-back configuration confirmed that we had wired the jacks properly. Next we temporarily cross-wired the dry copper circuit to a working phone line at the Observatory and then, using a lineman's butt-set, accessed that phone line from the other end of the dry copper circuit at Converse Hall. This confirmed that continuity existed between Converse and the Observatory.
With careful planning and diligent testing completed, all that remained before establishing a link between the Converse and Observatory HDSL modems was to select a common data rate that did not exceed the capability of the dry copper circuit - or so we thought. We started with 1152kb/sec – the highest selectable data rate for one-pair operation – but could not establish a link. We stepped the modems down in 64K increments all the way to 64kb/sec – the lowest selectable data rate for one-pair operation – but still could not establish a link. The problem turned out to be the type of circuit we had installed – a BANA circuit. According to Bell Atlantic, BANA circuits run through a remote monitoring system that was obviously interfering with our equipment. They replaced it with a PLNA circuit which, they say, is not remotely monitored. The initial link was established at a data rate of 1024kb/sec, but reliability was marginal. Stepping the modems down to 960kb/sec provided a stable link that has been maintained continuously since the end of January. Since the Campus-HRS Remote is an SDSL product (the “S” representing “symmetrical”), the 960kb/sec data rate is available in both the upstream and downstream directions.
In April we started to investigate the possibility of a second application of DSL over LEC dry copper. Another small academic department would soon be taking up temporary residence in an off-campus house while their on-campus building underwent major renovation. This department was accustomed to having “full ethernet” access to the campus data network and anything substantially less - like dial-up modem access - was going to feel woefully inadequate. The Bell Atlantic LMU results indicated an acceptable distance between this house and Converse Hall. Since this was to be the second of possibly several additional dry copper circuits we would be ordering from Bell Atlantic, it seemed like a good time to learn the correct USOC to specify on the service order. It stood to reason that placing an order with a USOC, rather than just a general description, should eliminate all of the guesswork involved in installing a circuit that worked for our application. It would take two months to learn the correct USOC.
Even though Bell Atlantic had, after several attempts, successfully installed and billed us for a dry copper circuit for the Observatory, they were not sure that the USOC that appeared on our invoice - 1LPBP - was correct. In fact, they said, it probably was not. They said a procedures manual did not exist that said “for customers seeking to deploy their own DSL service, provision a Type X circuit with a USOC of Y.” I perused a 10-year old copy of the 3-inch thick USOC manual, but it is very cryptic and I found it difficult to use. I surfed the web in search of an online version of the USOC manual, but did not find one. I was hoping to find one with a search engine that allowed me to specify my need in layman's terms – “a dry copper circuit with no loading coils and no bridge taps.” PairGain tried but could not provide any assistance. Finally we found a Bell Atlantic service order representative who was familiar with our request, having recently written an order for a nearby municipality that had a similar application. He said we needed something called an “intraexchange private line with no signaling” with a USOC of PVLFL. I quickly found this code in the USOC manual and confirmed the rep's description of the circuit. PVLFL was in a class of service called “DATA SVC – PVT LN OR COMPUTERIZED TRAFFIC CONTROL.” The description for PVL++ was “VOICE GRADE – USED WH CPE – 2000 SERIES.” The first suffix of F described the channel characteristics as “TYPE 2001 – VOICE GRADE – CUSTOMER PROVIDED EQUIPMENT.” The second suffix of L described the tariff jurisdiction as “INTRASTATE – INTRALATA – INTRAEXCHANGE.” Intraexchange (or within an exchange) indicates that only one central office exists between the two points. There was a long list of first and second suffixes that could be used to specify a desired circuit. I noticed a second suffix of S that had a description of “INTRASTATE – INTRALATA – INTEREXCHANGE.” Interexchange (or between exchanges) indicates to me that a PVLFS would be used to order a circuit when two or more central offices exist between the two points. Maybe the USOC manual wasn't so difficult to use after all.
Our second dry copper circuit – an RTNA circuit - was ordered and provisioned
with a USOC of PVLFL in late June. On July 2 it was installed, correctly
I might add, and on the next business day we established a DSL link with
a data rate of 1,152kb/sec. This proved, at least with our LEC, that
specifying the correct USOC is the most important ingredient for a successful