6/98 DataComm LAB TEST. DSL: Worth Its Wait
(positive review of ASND IDSL)
data.com
(excerpts below. See above link for entire article)
By David Newman, Michael Carter, and Helen Holzbaur
Easier, faster, cheaper: Those are just some of the advantages emerging DSL technologies hold over T1/E1 leased lines
TOP PERFORMERS
Time for a pop quiz. DSL technology is a) Highly
cost-effective as a T1/E1 replacement b)
Networking's newest source of inflated
performance claims c) A useful way to augment
leased lines d) Not much of an improvement over
current digital services.
Confused? No wonder. There are nine different
types of DSL floating around, and with many of
them aimed at residential users it isn't clear what
corporate customers stand to gain.
To bring things into sharper focus, Data Comm
teamed up with National Software Testing
Laboratories (NSTL, Conshohocken, Pa.) and
conducted the industry's first public comparison of
DSL (digital subscriber line) technology. Instead of
evaluating similar products in the same category,
we compared different DSL technologies to see
which fares best for corporate use.
Table 1: Selected
Vendors of DSL
End-User Devices
Vendors made our job easier by failing to supply
five of the nine DSL varieties. But three of them
were able to come through with the other four:
Ascend Communications Inc. (Alameda, Calif.)
supplied IDSL (ISDN-like DSL). Paradyne Corp.
(Largo, Fla.) delivered MSDSL (multirate
symmetric DSL) and RADSL (rate-adaptive
DSL). And Telmax Communications Corp.
(Fremont, Calif.) supplied products based on the
emerging HDSL 2 (high-bit-rate DSL version 2)
spec (see Tables 1 and 2).
Posting Results
Figure 1: Comparing
Relative DSL
Performance
Figure 2: Comparing
Absolute DSL
Performance
So how well did the DSL technologies stack up?
Because each tops out at a different rate, we've
normalized all results by presenting them as
percentages of theoretical maximum throughput
(see Figure 1). That's a fair way of comparing the
ability of each to make good on its potential. But
some readers are interested in raw performance
above all, so we've also presented results in terms
of absolute throughput (see Figure 2).
In comparing the normalized results, IDSL
(represented by Ascend's MAX DSL and Pipeline
50) appears to have a distinct advantage over the
rest. In the raw BLAST tests, it showed almost no
degradation in throughput for all but one
impairment. IDSL appears even more stable when
handling live application traffic: Results are virtually
identical, regardless of how bad line conditions
become. Latency was also rock-solid across all
impairments: 17 microseconds for 64-byte
payloads and 139 microseconds for 1,024-byte
payloads. These numbers suggest IDSL would be
a good choice for applications that demand very
predictable response times.
_____________________________________________________________________
Data Comm and NSTL gratefully acknowledge
the support of vendors that supplied equipment
for this test. Thanks especially to Ascend,
Paradyne, and Telmax for taking part in a
comparison of technologies rather than
products. Thanks also to Consultronics Ltd.
(Concord, Ontario), which supplied the
DSL-400 Wire Line Simulator and DLS-90
18-kfeet Extender, and to Network Associates
Inc. (NAI, Santa Clara, Calif.), which supplied
an Expert Sniffer protocol analyzer. Thanks
also go to Steven Eliot of McGraw-Hill
Professional Publishing, who supplied a copy of
ADSL and DSL Technologies by Walter
Goralski (McGraw-Hill, 1998), a useful
reference in researching this test.
David Newman is senior technology editor for
Data Comm. His e-mail address is
dnewman@data.com. Michael Carter is a
project manager at National Software Testing
Laboratories Inc. (NSTL, Conshohocken, Pa.);
his e-mail address is mikec@nstl.com. Helen
Holzbaur is manager of licensing and
methodologies for NSTL and can be reached at
helen@nstl.com. |
