The Need For Speed And The Copper Wire

Lisa A. Phifer
Core Competence, Inc.
lisa@corecom.com
Originally published by Interop OnLine, June 1997; reposted with permission.

You know the names: ISDN, K56Flex, x2, cable modems, ADSL, IDSL, xDSL.

The question is which of these technologies will win the remote access market? Industry players are scrambling to predict the answer. Big bucks are riding on this.

Just about the only thing we can know for certain about tomorrow's remote access market is that it will be vastly different - and a good bit faster - than today's. And though we can't predict the winner, we do know how each technology stacks up against its rivals right now.

ISDN

It's about time we changed the colloquial expansion of this acronym to "It Sure Does Network!" - because it finally does. ISDN may have taken 10 or 15 years to get off the ground, but it's here now and it works.

According to International Data Corp., there are 500,000 ISDN subscriber lines installed in the US, with a nine-fold increase anticipated by the year 2000. Bell Atlantic alone currently installs 4,000-5,000 new lines per month, and NYNEX reports a 74 percent increase in the number of subscriber lines installed during the past year. What accounts for this long-awaited market acceptance?

ISDN

ISDN modems, called terminal adapters, are now widely available in a variety of form factors through consumer retail channels for as little as $200. For small business applications, dozens of ISDN routers and bridges can be found, starting around $600. ISDN-based Internet access can be purchased from all but the tiniest "mom and pop" Internet Service Providers (ISPs). The cost of single-host ISDN Internet Access is competitive with analog dial. My own ISP, BellAtlantic.Net, charges just $3 more per month for unlimited 64 Kbps ISDN internet access versus 28.8 Kbps analog internet access.

2B+D Basic Rate ISDN offers two 64 Kbps B channels that can be bonded together to form a 128 Kbps "data pipe". ISDN has been demonstrated at our test lab to yield over 300 Kbps with compression. This second B channel also provides the flexibility to support simultaneous voice or fax calls without a separate analog line. And ISDN B channels are symmetric: they support the same bandwidth in both upstream and downstream directions, making ISDN attractive for video conferencing and remote LAN interconnection. Because ISDN is switched, it can also be used in conjunction with permanent high-speed access lines such as T1 and Frame Relay to provide backup or extra bandwidth. Finally, ISDN B channel call setup is at least 10 times faster than analog modems. It may soon support an "always on" 16K D channel for continuous, low-bandwidth data (can you say PointCast?)
Always On/Dynamic ISDN (AO/DI)

AO/DI utilizes the ISDN D channel to support a constant stream of low-bandwidth traffic, with automatic rollover to a B channel on demand -- perfect for "push" technology, email, point-of-purchase, banking, and other short message transactions. Pacific Bell, Southwestern Bell, and Bell South will have announced service offerings by the time you read this. Cisco, 3Com, USR, Shiva, Eicon, Intel, and Motorola are expected to ship AO/DI-compliant products before the end of 1997.

Ok, so why are we reading that ISDN stands to lose market share to newcomers 56K and cable modems? The reasons start with cost. ISDN requires a separate digital subscriber line to your home or office, either in addition to or instead of your existing analog line. Most telcos have tariffed ISDN as a metered service, typically charging from one to four cents per minute per B channel. And although telcos have long ago upgraded most central office switches to support ISDN, they must provision ISDN as a switched circuit. Provisioning may not mean much to you as a consumer. But it's a headache for telcos who would prefer to run data services separately from voice traffic. ISDN installation and configuration at the customer end is still a tad more difficult than for analog modems. And old vendor/switch/service incompatibility stories still haunt ISDN, despite vast improvements.

And, of course, in an industry crazed by "the need for speed", an ISDN BRI's 128 Kbps pales in comparison to the multi-megabit data rates touted for ADSL and cable modems. An ISDN upgrade to Primary Rate (PRI), with its 23 B channels, is a capable of supporting 1.54 Mbps. But PRIs are marketed to commercial enterprises and service providers, not to residential customers for whom it is prohibitively expensive.

x2 and K56Flex Modems

If 128 Kbps isn't enough, you ask, then why would anyone consider 56 Kbps? The answer is surprisingly simple: 56K modems operate over your existing analog line, are installed and configured just like older analog modem, can still interoperate with slower modems, and may even still work at 28.8 or (shudder!) 14.4 with your current Internet Access account. K56Flex  x2

56K modems are hitting the consumer retail shelves at under $200. Some vendors are even offering to upgrade 33.6 Kbps modems to 56 Kbps at little or no cost. And new PCs from Compaq and IBM are already shipping with 56K modems. With a deal this sweet, how can you resist?

The answer can be summed up in a single word -- interoperability. Or lack thereof. 56K modem vendors have broken down into two incompatible camps: x2, lead by USR, and K56Flex, lead by Rockwell and Lucent. These camps went to war in the standards arena to forge a common ITU V.PCM standard that will provide the needed interoperability but require upgrades to every 56K modem purchased before the standard is finalized. In the interim, vendors and ISPs alike are forced to take sides - or double their cost (see chart). As a consumer, it is imperative that you match your analog vendor to your ISP, or you won't be able to exceed plain old 28.8 or 33.6 with your new 56K modem.

And while we're on the subject, let's talk about speed for a minute. You might expect a "56K" modem to operate at 56 Kbps -- but it won't. The technology which squeezes more speed out of analog over copper is actually limited to 53 Kbps downstream (web pages, downloaded files) and 33.6 upstream (outgoing email). Moreover, this technology is extremely sensitive to the "noise" found on analog lines. You may connect at speeds in the 40s, or, in extreme cases, you may not be able to connect at all. Vendors such as USR are providing 30-day money-back guarantees to encourage you to "try before you buy" rather than be dissuaded by this problem.

56K requires digital terminations at the head-end and specialized modem pools that not only require investment now (USR quotes $60-95 per port), but also an upgrade once the V.PCM standard is final. About 600 ISPs have been driven by consumer demand to upgrade their infrastructure to support 56K (see chart), while others have adopted a "wait and see" attitude. Those ISPs supporting 56K access are likely to pass along the added infrastructure cost to customers as a surcharge ($10 extra per month at Netcom). Expect ISPs to break their knuckles on 56K while enterprises wait for the technology to mature before adopting it for remote access.

56K Chart  56K Chart

Cable Modems

While ISDN is here today and 56K is here this year, the same cannot quite be said for cable modems. Numerous trials are underway, but most industry analysts agree that we won't see a real cable modem rollout until 1998 or 1999. What has everyone so excited? Cable Modem

Cable service providers seek to leverage from the estimated 100 million homes now accessible by coaxial cable. By divvying up the RF spectrum available over coax, cable providers believe they can bring multi-megabit downstream data -- one to 40 Mbps -- to your set-top box, TV, or an interface card on your PC. Even at 1.2 Mbps -- the rate supported by Scientific Atlanta's dataXcellerator modem -- that's still four times faster than two ISDN B channels, bonded with compression. And it's more than 200 times faster than 56K modems!

Ready to subscribe? Let's dig a little bit deeper first. The ISP charge for cable modem service is running double the charge for ISDN or 56K analog unlimited Internet access. Where I live, Comcast@Home trials charge $40 to 50 per month, including modem rental. This is, of course, in addition to charges to cable TV and analog telephone service. Unlike ISDN or 56K, cable modems won't initially support voice traffic.

What drives up the cost of cable modem service? Cable modems still cost about $500 each, two to three times a 56K modem or ISDN TA. Typically, cable operators buy them in volume and lease them to you. At the same time, they have to finance layinga check of a lot of new cable. Cable operators must build fiber networks to carry multi-megabit traffic to hundreds of homes, and add two-way transmission gear to their largely one-way distribution plant. Industry analysts estimate that only 5 to 15% of the 100 million home now accessible by cable are two-way capable.

Since providing this upstream bandwidth over cable would prove so costly -- and would be limited to hundreds of Kbps at best, anyway -- several manufacturers have chosen a different path altogether: plain old copper. Intel's CablePort and Scientific Atlanta's dataXcellerator modems use a telco-provided analog line for upstream, and Motorola has indicated plans to follow suit. So, better factor in the cost of a second analog line…You get the picture.

Cable modems also currently suffer from the same incompatibility that plagues 56K. The recent MCNS standard, published in March 1997, provides interoperability at the MAC layer, but says nothing about what rides over this raw data pipe. Throw in another competing standard -- IEEE 802.14 -- and you have a mess that will take vendors at least another year or so to overcome. For the immediate future, expect to lease your modem from your cable operator and use only their online services.

Despite all of this, cable modems probably do have a rosy future in the residential market -- it just isn't here yet. If cable modem downstream speeds blow the doors off 56K and even ISDN, where will the competition come from?

ADSL

Don't expect telcos to sit and watch quietly while cable operators ramp up multi-megabit data services. They've already fired what they hope is a preemptive strike with Asymmetric Digital Subscriber Line (ADSL) and its many-lettered cousins -- IDSL, SDSL, HDSL, VDSL -- collectively known as xDSL.

ADSL

Like 56K and ISDN, ADSL runs over plain old copper. That's why the telcos love it so. Initial deployment boasts speeds of up to 1.544 Mbps downstream and 64 Kbps upstream. And its available on a continuous basis; unlike 56K and ISDN, ADSL is not switched. At these speeds, ADSL compares favorably to today's cable modems, with additional bandwidth promised to follow. ADSL is easier for telcos to provision than ISDN, too.

You'd think this would translate into lower costs, but it hasn't yet. ADSL modems even more costly than cable modems -- typically $1-2K -- resulting in a recurring monthly subscriber cost of $50 to 200 per month.

Because ADSL is point-to-point, it does not enable mobility like 56K or ISDN. The digital subscriber loop runs from your home or office to the telco's central office, and data services must terminate at a fixed endpoint. This positions ADSL as a candidate for enterprise LAN connectivity, interconnecting remote and branch offices to corporate headquarters or an ISP. But if the asymmetric nature of ADSL gets in the way, you'll need to consider HDSL or SDSL instead.

Not to be outdone by the infighting within the 56K and cable modem camps, ADSL has its own share of interoperability issues -- principally a battle between two modulation techniques known as DMT and CAP -- as well as market fragmentation caused by the proliferation of xDSLs. Sound familiar? Think incompatiable ISDN vendor-protocol variations, 5 or 10 years ago.

But I'm optimistic about DSL -- a year or two down the road. Trials are underway by NYNEX, Pacific Bell, Southwest Bell, Ameritech, Bell Atlantic, GTE, US West, and Bell South - these folks have much deeper pockets the your typical cable operator. And they are highly motivated to beat the cable industry -- particularly in the business market. Dataquest predicts 3.7 million ADSL subscribers by the year 2000, and that's just the tip of the iceberg.

From 60-70% of U.S. homes are close enough to a telco central office to make use of ADSL. You'll often see 18,000 feet specified as the distance limitation for ADSL, but the real limit is a factor of the telco infrastructure (line splices and taps) between you and your central office. Poor line quality chips away at the top speed that can be achieved over your subscriber line. But, heck, wouldn't you be happy with 800 Kbps? 600 Kbps?

To narrow the price/performance gap between ISDN and ADSL, vendors such as Copper Mountain offer lower-cost SDSL gear at 384 Kbps (bi-directional) or 768 Kbps (asymmetric). Ascend took another tack by creating IDSL, a combination of ISDN and xDSL technology that achieves no better than ISDN BRI throughput, but at a flat rate recurring cost - essentially an "entry level" xDSL. And the list goes on, as telcos seek to boost data speeds over copper wire.

How Do You Choose?

Anyone hoping for a simple answer will be disappointed. However, try answering two simple questions: Why do you need remote access? And how much speed is enough?

It's easy to compare upstream and downstream throughput for these remote access technologies, so let's put them into perspective. Consider the time required to transfer 1 MB of data.

Transfer Rate

Transfer Rate

If you intend to use your remote access primarily for downloads from the Internet, transfer rates arguably peak somewhere between 128 Kbps and 500 Kbps (63 to 16 seconds per MB). And they vary greatly, affected by a multitude of factors. Now take a second look at the chart, drawing a line somewhere in this range, depending upon your own experience, and see how that changes the outcome.

If you intend to use your remote access primarily to connect with a remote LAN, focus on the upstream bandwidth. You'll either be constrained by this limit or need to work your way around it -- by purchasing two asymmetric pipes, for example.

Are you a home user, a telecommuter, small business, or other enterprise?

The average home user is primarily interested in Internet or online service and transfers about 10 times more data downstream than upstream. For the low-end home user, 56K modems are the most logical immediate upgrade choice. The power home user should consider ISDN, an attractive though more expensive alternative, and cable modems when they become available in the area.

The telecommuter, on the other hand, may be wise use ISDN right now. This crowd needs the increased productivity and efficiency that higher bandwidth and multiple data channels bring, along with the flexibility of switched access. If possible, choose an ISDN TA or router that supports analog dial as well, so you'll have the best of both worlds (especially when traveling). Telecommuters can dream about higher-speed access, but it'll take a year or two and a drop in price before the cost of xDSL can be justified for most telecommuters.

Small offices, look at your bandwidth and usage requirements. ISDN may make perfect sense for remote LAN connectivity to either the corporate enterprise or the Internet. If your online hours or bandwidth requirements exceed the capacity of ISDN, you've probably already considered a leased line, Frame Relay, or SMDS -- and you should take a serious look at xDSL trials in your area. Forget cable modems. They are targeted squarely for the residential market. Even after you move to xDSL, you may still want to consider ISDN for backup or bandwidth-on-demand. And choose carefully between xDSLs -- factor cost, both upstream and downstream traffic patterns, and anticipated growth.

So, make an informed guess and jump right in. There are plenty of products out there to quench your thirst for speed!