Remote Access with Modems

When discussing the limitations of serial signaling, it would be remiss not to discuss the limitations
of the public phone system and the analog technology available today. As noted earlier
in this chapter, asynchronous connections are limited to 56Kbps, or 53Kbps by FCC order. Distance
and line quality further limit this amount of bandwidth, possibly reducing throughput to
28.8Kbps or less. (This was written in a Boston hotel room, where there was no reliable connection
beyond 26Kbps.) In addition, connections might take up to a minute to establish and
might be further impeded by load coils and analog-to-digital conversions between you and the
central office (CO). Load coils are amplifiers used to accommodate longer distances than normal,
and analog-to-digital conversions are often used in new housing developments to convert
the copper pairs to fiber, again extending the length of the link. It is far cheaper to run a few
pairs of fiber to an access terminal (a small cabinet that sits in the neighborhood and converts
the fiber to copper) where the copper runs than it is to extend into the home.
You need to remember that the plain old telephone service (POTS) is exactly that—old. It was
developed from the same technology that Alexander Graham Bell developed in his lab over a
hundred years ago and was never intended to address the needs of video and data. That’s the
first problem with analog connections: they were never designed to allow millions of bits of
data to flow from one point to another.
The second problem with analog connections is their inefficiency. Voice is a specific type of
data and fits in a single 64Kbps channel. You might already be aware of the channels of voice
aggregation, or T-1 circuits—where 24 voice signals (DS0, digital signal) fit into a T-1 or DS1.
Data is unlike voice, however, which leads to inefficiency. Voice demands that the idle (or no
data) points in the conversation be communicated as well, so there is always a constant flow
of information. Data doesn’t work that way; if no data is transmitted, there is little need for the
bandwidth to be consumed. By using only the available bandwidth that is necessary, it is possible
to service more connections with data than voice. You might have heard of convergence
or time division multiplexing (TDM), two very different concepts that relate to this topic.
Convergence
is the concept of voice, video, and data all using the same network, whereas
TDM
is
the old voice channel model—each channel always given the same amount of access to the network
regardless of the need. Convergence will remove TDM from the network and place everything
into packets that can then use only the required amount of bandwidth, as opposed to
reserving more than is necessary.
However, convergence will also effectively eliminate the analog network (an event that has
already occurred in the core of the telephone world). But before that comes to fruition, network
administrators will need to contend with the problems of the current network, including long
call-setup times, poor-quality connections, and low bandwidths.
These problems, just for the record, already have solutions in many cases. Although it is true that
analog connections are the most prevalent in the world, the availability of DSL, cable, ISDN,
Frame Relay, wireless, and Long Reach Ethernet (LRE) enables designers to incorporate alternatives
into their installations and provides an indication of what will happen in the near future.
At the beginning of this sidebar, we noted problems with analog service and the phone network.
While discussing these problems, we failed to address what is possibly the most important
problem—cost. Readers of the
CCDP: Cisco Internetwork Design Study Guide
(Sybex,
2000) will recall the emphasis on business concerns when designing the network. Cost is frequently
the single biggest business factor, period. Business managers who do not understand
bits and protocols certainly understand the benefits of a $40-a-month fixed cost per employee
compared to a variable bill that could surpass $100 a month.
One last item: virtual private networks.
Virtual private networks (VPNs)
are encrypted sessions
between two devices over the public network, typically the Internet. These sessions are virtually
private because the encrypted data is, conceptually, protected from snooping. Users, however,
will still be affected by delay and bandwidth limitations that could be better controlled in
private network installations.
VPNs provide remote access designers with two benefits. The first is low cost, which, as noted in
the previous paragraph, is a powerful business case argument. The second benefit is universality—
or the capability to allow access from different technologies. With VPNs, the administrator no
longer cares what technology is used on the remote side of the connection. The remote side simply
needs to connect to the Internet via any available transport, or in some cases, an internationally
accessible single-vendor network (which can provide service-level agreements and other service
guarantees). Once connected, the connection traverses the network and is decrypted at the corporate
access point, typically a T-1 or DS3, depending on the bandwidth demands. For smaller support
departments, this entire service might be outsourced so the maintenance of the VPN
equipment and connections is not an additional burden on the team.