Cisco AG241 Bedienungsanleitung PDF herunterladen (Seite 3)

apcmag.com  june 06  apc labs

apc labs  june 06  apcmag.com

softwaresoftware

2006

JUNEDETAILED PRODUCT REVIEWS AND COMPARISONS

REVIEWS

BEST BUYS

LABS CHALLENGE

labs

apcmag.com  june 06  apc labs

apc labs  june 06  apcmag.com

softwaresoftware

2006

JUNEDETAILED PRODUCT REVIEWS AND COMPARISONS

REVIEWS

BEST BUYS

LABS CHALLENGE

labs

welve months ago, when APC

embarked on Australia’s first

published lab testing of ADSL2+

modems, we weren’t prepared for

the enormous variation in results

that we’d see.

Some modems could sync at

ADSL2+ speeds (up to 24Mbit/s),

but didn’t have an adequately

grunty CPU to chew through

data faster than about 12Mbit/s.

Confusingly, the best-performing

modem, Billion 7402, used the same

processor (Conexant) as one of the

poorest (Draytek 2800).

This year, we expanded our

testing to cover two distinct

categories of modem/router:

1. Budget models that only include

the essentials: an ADSL2+ modem/

router and four ethernet ports.

2. “All-in-ones” including an

802.11g access point, VoIP ports

and (in some models) VPN access

to allow a secure, direct connection

to your home network from a

remote location.

We also added wireless

performance testing to our range

of tests. With 802.11g’s “54Mbit/s”

performing more like 20Mbit/s in

optimal conditions (and that speed

dropping away rapidly as you move

away from the base station), we

wanted to test our concerns that

wireless would lead to a bottleneck

in any ADSL2+ connected home

network. Unlike other publications

— that have simply plugged

modems into a home ADSL2+

connection of unknown line length

and condition — we used the test

lab of ISP Internode, with assistance

from their R&D department and

network experts. We had full access

to their DSLAM configuration

interface to monitor modem

performance in detail.

HOW WE TESTED

There are two ways to test ADSL2+

modems: using an electronic phone

line simulator which can be configured

to inject variable amounts of noise

into a line, or using real lengths of real

copper wire. Though both methods

have their pros and cons, we opted to

test with real copper wire (the closest

thing to what actually runs between

your home and a Telstra exchange).

Of course, because the copper

wire we were testing on was new and

good quality, our results are probably

a little better than what would be seen

in the field. Real Telstra phone lines

have to contend with moisture in pits,

bridge taps (bits of unused copper

wire hanging off your line) and a

multitude of other problems inherent

in a decaying 100-year-old network.

However, our tests were designed

to give an accurate comparison of how

different modems perform over

various line lengths. The longer the

line, the greater the background

noise, as outside signals are picked

up from nearby lines and radio

frequency interference. In fact,

because the copper phone line we

tested with was batched in enormous

coils, the cross-talk from the hundreds

of nearby strands would have gone

some of the way to simulating what

happens in a Telstra main trunk line:

signals leak within the fat batch of

cables that runs underground from the

exchange before being split up to

customers’ homes.

We tested at three line-lengths:

1km, 3.5km and 6km. We had access

to a “Krone block” where it was

possible to change the line length

in 500m blocks by plugging in at a

certain jumper position.

We were also able to monitor

modem sync speed and line

conditions via the Ericsson PEM

configuration software which

displayed details on the Ericsson

EDN312xe DSLAM (that was

pumping signal down the copper

line to our modems).

We downloaded large files from a

dedicated FTP server running FreeBSD

to determine actual throughput.

The server had 2GB RAM and an

AMD Athlon 2200XP processor

(and was commissioned especially

for APC’s testing with no other users).

For the wireless testing, we

tested all the modems at APC’s

Sydney CBD offices using the

established test locations that we use

for other wireless router testing. Our

offices are far from a “perfect” test

environment (there are many wireless

networks in nearby buildings and

several in use on our floor). However,

rather than providing “perfect lab

conditions” for wireless, we wanted

to get an idea of what constitutes

real-life performance for the modems.

We conducted a site survey

using NetStumbler and confirmed

that an unused wireless channel

was available. We configured all

the modems to use that channel

before testing.

The speed of an unencrypted

wireless signal is really only useful for

hotspot operators to know — home

and business users want secure

wireless networks. With that in mind,

we tested with WPA-PSK turned on

— the best level of security

commonly available across all the

modems. So, in addition to testing

wireless signal penetration and

throughput, this tested the modem

CPU’s ability to encrypt data at

high speed.

We used the IPerf software

running on a server connected by

Ethernet to each modem to pump out

a stream of TCP data as fast as the

modem could take it. Our client

machine was a Centrino-compliant

Sony Vaio TX17 notebook with all

power-saving features switched off.

We tested the wireless

performance in three positions: about

30cm from the transmitter, about

10m away (through a wood partition,

a plaster wall and a glass wall) and

about 15m away (through wood,

plaster, metal partitions, racks of

magazines, photocopiers and other

office equipment).

This Krone block allowed

us to change the length

of the test copper phone

line in 500m increments.

The actual Ericsson DSLAMs we

tested with in Internode’s lab.

This high speed FreeBSD FTP

server was dedicated to APC’s

testing — to ensure server load

wasn’t an issue.

The back of the DSLAM

where copper twisted-pair

meets high-speed

digital signals.

1 2 3 4 5 6 7 8 ... 13 14

Kommentare zu diesen Handbüchern

Keine Kommentare

Cisco AG241 Bedienungsanleitung Seite 3

Kommentare zu diesen Handbüchern

Verwandte Produkte und Handbücher für Vernetzung Cisco AG241