Fastest Router possible? Broadband Router Reviews?

[Bruce B]

I have 3mb ADSL from my service provider, and I want to make sure I am
getting the most out of my connection.

I'm mostly using the connection for gaming.

Can someone recommend the fastest router available?
I currently have a D-Link 704 revision B. I read that revision C has a
faster processor and was wondering if it would be worth buying the new one,
or is there perhaps a better one available from another manufacturer? I
have been fond of netgear hardware in the past.

Also, does anyone know of a site with compartive router reviews?

 

[Jim Turner]

I have 3mb ADSL from my service provider, and I want to make sure I am
getting the most out of my connection.

I'm mostly using the connection for gaming.

Can someone recommend the fastest router available?
I currently have a D-Link 704 revision B. I read that revision C has a
faster processor and was wondering if it would be worth buying the new one,
or is there perhaps a better one available from another manufacturer? I
have been fond of netgear hardware in the past.

Also, does anyone know of a site with compartive router reviews?

I think you will find that you are not limited by the router, even for
gaming. Your internet connection will be limited by what your ISP can
provide, and on the local side, you should have a 10/100 switch. Faster
processors only make sense in a router if you are having them block a lot
of sites, etc.. If you are doing more advanced content filtering, a
faster processor can be useful For games, turn off filtering.

I like netgear, but your Dlink router should be fine unless you need
advanced features.

 

[kony]

I have 3mb ADSL from my service provider, and I want to make sure I am
getting the most out of my connection.

I'm mostly using the connection for gaming.

Can someone recommend the fastest router available?
I currently have a D-Link 704 revision B. I read that revision C has a
faster processor and was wondering if it would be worth buying the new one,
or is there perhaps a better one available from another manufacturer? I
have been fond of netgear hardware in the past.

Also, does anyone know of a site with compartive router reviews?

You need to ping "something", while going through the router, a few
times to get a good average ping time, then immediately (proceed to
change the connection) connect without the router, directly through
your gaming system. Now get another average ping time and compare
them. I doubt that you'd see substantial enough performance increase
to bother upgrading one revision with the same product.

I have a D-Link 704, don't know the revision number but it's probably
not the revision C, is back when they still had a metal casing, and it
adds about 5-9 ms.

Probably the fastest pings would be seen by using an internal PCI ADSL
card.


Dave

 

[Stacey]

I have 3mb ADSL from my service provider, and I want to make sure I am
getting the most out of my connection.

I'm mostly using the connection for gaming.

Can someone recommend the fastest router available?
I currently have a D-Link 704 revision B. I read that revision C has a
faster processor and was wondering if it would be worth buying the new
one,
or is there perhaps a better one available from another manufacturer? I
have been fond of netgear hardware in the past.

Also, does anyone know of a site with compartive router reviews?

I just tesed my "router" which is a pentium 233 (Oclocked to 266) running
Mandrake 8.2 with two 10/100 nics set up as a firewall/router. I pinged
several sites from that system, which is directly connected to an alcatel
speed touch home ethernet DSL, and then from systems on my network and get
the same ping time averages. Pinging that box from anywhere on the local
network give an avg of .38ms so I assume that's all it adds to my ping
times? As I said I see no difference in the averages from that system and
boxes on the network. I'm pretty sure a P233 is "hotter" than anything a
hardware router has in it! Should be able to find a pentium 1/2 system
cheaper than any router and can configure it to your hearts content. The
other advantage is you can also set it up as a caching nameserver which
saves lookup times.

 

[Gerhard Beulke]

I just tesed my "router" which is a pentium 233 (Oclocked to 266) running
Mandrake 8.2 with two 10/100 nics set up as a firewall/router. I pinged
several sites from that system, which is directly connected to an alcatel
speed touch home ethernet DSL, and then from systems on my network and get
the same ping time averages. Pinging that box from anywhere on the local
network give an avg of .38ms so I assume that's all it adds to my ping
times? As I said I see no difference in the averages from that system and
boxes on the network. I'm pretty sure a P233 is "hotter" than anything a
hardware router has in it!

Of course since it has fifteen times the power consumption as a
"normal" hardware router.
;-)

Should be able to find a pentium 1/2 system
cheaper than any router and can configure it to your hearts content. The
other advantage is you can also set it up as a caching nameserver which
saves lookup times.

Routers are dirt cheap now. Why bother with an old crappy PC and waste
your time setting it up proper?
And the routers have the modem built in...
The power savings alone pay the hardware router off after 4 months
continuously run of the crappy PC system acting as the router.

Cheers

 

[kony]

I just tesed my "router" which is a pentium 233 (Oclocked to 266) running
Mandrake 8.2 with two 10/100 nics set up as a firewall/router. I pinged
several sites from that system, which is directly connected to an alcatel
speed touch home ethernet DSL, and then from systems on my network and get
the same ping time averages. Pinging that box from anywhere on the local
network give an avg of .38ms so I assume that's all it adds to my ping
times? As I said I see no difference in the averages from that system and
boxes on the network. I'm pretty sure a P233 is "hotter" than anything a
hardware router has in it! Should be able to find a pentium 1/2 system
cheaper than any router and can configure it to your hearts content. The
other advantage is you can also set it up as a caching nameserver which
saves lookup times.

38ms maybe, but 0.38ms is likely impossible on the fastest _PC_ ever
built.


Dave

 

[Stacey]

38ms maybe, but 0.38ms is likely impossible on the fastest _PC_ ever
built.


Dave

[stephe@friend stephe]$ ping 192.168.0.1
PING 192.168.0.1 (192.168.0.1) 56(84) bytes of data.
64 bytes from 192.168.0.1: icmp_seq=1 ttl=255 time=0.449 ms
64 bytes from 192.168.0.1: icmp_seq=2 ttl=255 time=0.394 ms
64 bytes from 192.168.0.1: icmp_seq=3 ttl=255 time=0.406 ms
64 bytes from 192.168.0.1: icmp_seq=4 ttl=255 time=0.389 ms
64 bytes from 192.168.0.1: icmp_seq=5 ttl=255 time=0.392 ms
64 bytes from 192.168.0.1: icmp_seq=6 ttl=255 time=0.423 ms
64 bytes from 192.168.0.1: icmp_seq=7 ttl=255 time=0.413 ms
64 bytes from 192.168.0.1: icmp_seq=8 ttl=255 time=0.392 ms

--- 192.168.0.1 ping statistics ---
8 packets transmitted, 8 received, 0% packet loss, time 7007ms
rtt min/avg/max/mdev = 0.389/0.407/0.449/0.024 ms

Unless I'm reading this wrong? This was a ping from 192.168.0.2 to the
gateway box.

 

[Stacey]

Routers are dirt cheap now. Why bother with an old crappy PC and waste
your time setting it up proper?

Because the computer was free?

And the routers have the modem built in...

?? What inexpencive router has a DSL modem in it?

The power savings alone pay the hardware router off after 4 months
continuously run of the crappy PC system acting as the router.

??? You really think it costs $25 a month to run an old PC? My whole power
bill in the winter is less than $40 for 4 computers and the rest of the
house!

 

[Stacey]

kony wrote:

38ms maybe, but 0.38ms is likely impossible on the fastest _PC_ ever
built.


Dave

ping bellsouth.net
PING bellsouth.net (216.77.188.40) 56(84) bytes of data.
64 bytes from filter.bellsouth.net (216.77.188.40): icmp_seq=1 ttl=247
time=26.2 ms
64 bytes from filter.bellsouth.net (216.77.188.40): icmp_seq=2 ttl=247
time=20.9 ms
64 bytes from filter.bellsouth.net (216.77.188.40): icmp_seq=3 ttl=247
time=27.7 ms
64 bytes from filter.bellsouth.net (216.77.188.40): icmp_seq=4 ttl=247
time=21.7 ms
64 bytes from filter.bellsouth.net (216.77.188.40): icmp_seq=5 ttl=247
time=18.7 ms
64 bytes from filter.bellsouth.net (216.77.188.40): icmp_seq=6 ttl=247
time=22.6 ms
64 bytes from filter.bellsouth.net (216.77.188.40): icmp_seq=7 ttl=247
time=27.9 ms

--- bellsouth.net ping statistics ---
7 packets transmitted, 7 received, 0% packet loss, time 6068ms
rtt min/avg/max/mdev = 18.766/23.732/27.969/3.350 ms

Adding 9ms would add almost 50% to the ping times wouldn't it?

 

[Phrederik]

Routers are dirt cheap now. Why bother with an old crappy PC and waste

Because the computer was free?

And what is your time worth? A router just plugs in and goes. A PC takes
TIME to configure AND maintain.

?? What inexpencive router has a DSL modem in it?

No router I know of has anything like a modem built in. Not sure what kind
of crack they were smoking when they said there was a modem built in.

??? You really think it costs $25 a month to run an old PC? My whole power
bill in the winter is less than $40 for 4 computers and the rest of the
house!

No matter what, a router will use less than 10% of the power of a PC. It may
not be a big concern, but it all adds up.

 

[kony]

kony wrote:



ping bellsouth.net
PING bellsouth.net (216.77.188.40) 56(84) bytes of data.
64 bytes from filter.bellsouth.net (216.77.188.40): icmp_seq=1 ttl=247
time=26.2 ms
64 bytes from filter.bellsouth.net (216.77.188.40): icmp_seq=2 ttl=247
time=20.9 ms
64 bytes from filter.bellsouth.net (216.77.188.40): icmp_seq=3 ttl=247
time=27.7 ms
64 bytes from filter.bellsouth.net (216.77.188.40): icmp_seq=4 ttl=247
time=21.7 ms
64 bytes from filter.bellsouth.net (216.77.188.40): icmp_seq=5 ttl=247
time=18.7 ms
64 bytes from filter.bellsouth.net (216.77.188.40): icmp_seq=6 ttl=247
time=22.6 ms
64 bytes from filter.bellsouth.net (216.77.188.40): icmp_seq=7 ttl=247
time=27.9 ms

--- bellsouth.net ping statistics ---
7 packets transmitted, 7 received, 0% packet loss, time 6068ms
rtt min/avg/max/mdev = 18.766/23.732/27.969/3.350 ms

Adding 9ms would add almost 50% to the ping times wouldn't it?

It looks that way, but referring to the previous post, Just pinging
the router box may not be very telling, as your previous times
indicated. Not knowing how many nodes away bellsouth.net is from
you, it also reveals little. The response times will vary for
everyone, so the only way to compare is on your end, with and without
the router-box in place.

If you take a traceroute time average to the first node though the
router and without, take the difference. It may be less than 9ms...
due to various bug crawling the 'net recently there's more congestion
than usual, would skew the results more as you work towards main
pipes.


Dave

 

[Bruce B]

(All) Thanks for the help so far.
My ping (Using 3MBit ADSL + DLink 704 router) averages 32 ms.

I haven't tested it straight/without the router or with a different one, but
I'd sure like to try it and see.

 

[Stacey]

It looks that way, but referring to the previous post, Just pinging
the router box may not be very telling, as your previous times
indicated. Not knowing how many nodes away bellsouth.net is from
you, it also reveals little.

It's close (it's my ISP) and why I used them to try to avoid internet
traffic being a part of the results. Pinging yahoo gives slower times as it
goes through more hops to get there. Not much I can do about that.

[stephe@friend stephe]$ ping yahoo.com
PING yahoo.com (66.218.71.198) 56(84) bytes of data.
64 bytes from w1.rc.vip.scd.yahoo.com (66.218.71.198): icmp_seq=1 ttl=239
time=77.4 ms
64 bytes from w1.rc.vip.scd.yahoo.com (66.218.71.198): icmp_seq=2 ttl=239
time=74.4 ms
64 bytes from w1.rc.vip.scd.yahoo.com (66.218.71.198): icmp_seq=3 ttl=239
time=75.1 ms
64 bytes from w1.rc.vip.scd.yahoo.com (66.218.71.198): icmp_seq=4 ttl=239
time=76.0 ms
64 bytes from w1.rc.vip.scd.yahoo.com (66.218.71.198): icmp_seq=5 ttl=239
time=75.2 ms
64 bytes from w1.rc.vip.scd.yahoo.com (66.218.71.198): icmp_seq=6 ttl=239
time=73.9 ms
64 bytes from w1.rc.vip.scd.yahoo.com (66.218.71.198): icmp_seq=7 ttl=239
time=74.6 ms

--- yahoo.com ping statistics ---
8 packets transmitted, 7 received, 12% packet loss, time 7068ms
rtt min/avg/max/mdev = 73.998/75.305/77.495/1.132 ms

This is the -first- node traceroute shows and hence the low ping times
shown.

[stephe@friend stephe]$ ping 209.149.96.97
PING 209.149.96.97 (209.149.96.97) 56(84) bytes of data.
64 bytes from 209.149.96.97: icmp_seq=1 ttl=253 time=13.2 ms
64 bytes from 209.149.96.97: icmp_seq=2 ttl=253 time=11.7 ms
64 bytes from 209.149.96.97: icmp_seq=3 ttl=253 time=12.4 ms
64 bytes from 209.149.96.97: icmp_seq=4 ttl=253 time=12.5 ms
64 bytes from 209.149.96.97: icmp_seq=5 ttl=253 time=13.7 ms
64 bytes from 209.149.96.97: icmp_seq=6 ttl=253 time=13.2 ms

--- 209.149.96.97 ping statistics ---
6 packets transmitted, 6 received, 0% packet loss, time 5053ms
rtt min/avg/max/mdev = 11.731/12.824/13.772/0.670 ms

This was done from the ruoter itself.

[stephe@localhost stephe]$ ping 209.149.96.97
PING 209.149.96.97 (209.149.96.97) from 68.158.78.245 : 56(84) bytes of
data.
64 bytes from 209.149.96.97: icmp_seq=0 ttl=254 time=12.298 msec
64 bytes from 209.149.96.97: icmp_seq=1 ttl=254 time=12.245 msec
64 bytes from 209.149.96.97: icmp_seq=2 ttl=254 time=11.315 msec
64 bytes from 209.149.96.97: icmp_seq=3 ttl=254 time=11.121 msec
64 bytes from 209.149.96.97: icmp_seq=4 ttl=254 time=11.902 msec
64 bytes from 209.149.96.97: icmp_seq=5 ttl=254 time=12.677 msec
64 bytes from 209.149.96.97: icmp_seq=6 ttl=254 time=12.970 msec
64 bytes from 209.149.96.97: icmp_seq=7 ttl=254 time=12.025 msec
64 bytes from 209.149.96.97: icmp_seq=8 ttl=254 time=11.838 msec
64 bytes from 209.149.96.97: icmp_seq=9 ttl=254 time=12.364 msec

--- 209.149.96.97 ping statistics ---
10 packets transmitted, 10 packets received, 0% packet loss
round-trip min/avg/max/mdev = 11.121/12.075/12.970/0.549 ms


I get almost the same results doing this on the gateway/router box itself so
there is no loss to speak of?

The response times will vary for
everyone, so the only way to compare is on your end, with and without
the router-box in place.

Maybe I'm confused but what I posted was I tested the ping times to
bellsouth.net directly from the gateway/router box and then from the
network through that gateway and saw basically the same average ping times.
More dependant on each ping than anything elese. There was no loss to speak
of, maybe .5ms either way depending on when I did it?

If you take a traceroute time average to the first node though the
router and without, take the difference. It may be less than 9ms...

Like I said earlier, I see no difference in the average and why I assumed
the only difference/loss would be the time it takes to go from the network
to the router which was .4ms or so. Also wouldn't traceroute show the
delay in the router? Again I'm not a networking pro and maybe I'm reading
the results wrong? If what I'm getting is true pass through times and the
hardware routers are causing a 9-10ms delay then the old computer for a
router is a much faster solution.

 

[kony]

I get almost the same results doing this on the gateway/router box itself so
there is no loss to speak of?

You might be getting a close approximation but it would only be by
coincidence, since the performance of the router is the issue. The
test can't be done from the router, because the router is exactly what
you want to remove, completely, from the equation.

Maybe I'm confused but what I posted was I tested the ping times to
bellsouth.net directly from the gateway/router box and then from the
network through that gateway and saw basically the same average ping times.
More dependant on each ping than anything elese. There was no loss to speak
of, maybe .5ms either way depending on when I did it?

With the router involved in any part of the ping or tracert, it's
speed is still part of the resulting times... You have to remove it,
but then the issue is, you'd have a time but couldn't necessarily
compare it accurately to someone else's [ times on _their_ lan and
then even more variance possible the futher the ping goes]. It's a
test only applicable to the same environment with the only change
being substitution of a different router or elimination of the router.

Like I said earlier, I see no difference in the average and why I assumed
the only difference/loss would be the time it takes to go from the network
to the router which was .4ms or so. Also wouldn't traceroute show the
delay in the router? Again I'm not a networking pro and maybe I'm reading
the results wrong? If what I'm getting is true pass through times and the
hardware routers are causing a 9-10ms delay then the old computer for a
router is a much faster solution.

The old-computer-router can respond to the ping but then take
significant time in processing the request and forwarding it, part of
the time reported till the next node's response. Your old system
might be around the same speed as a hardware router, maybe faster, but
if I were to take a random guess it'd be that it's much slower due to
more overhead in the busses, the OS. I used to have an
old-system-router and wish i remembered the ping times but I do recall
that upping the PCI and memory bus had as much of an effect as CPU
speed (within limits), which is unfortunate since even today's sytems
are still stuck behind a 33MHz PCI bus unless you devote a server
chipset/board to it.

On the other hand, depending on how you're configuring your router it
could be doing time-saving things as well, like DNS caching. That
could make the difference.

I used to have a Freesco router box but I don't care about minor ping
variances so once I started using a hardware router I never looked
back. Since the hardware router does use less electricity I have more
uptime on an UPS if/when the power fails.


Dave

 

[Stacey]

The old-computer-router can respond to the ping but then take
significant time in processing the request and forwarding it, part of
the time reported till the next node's response. Your old system
might be around the same speed as a hardware router, maybe faster, but
if I were to take a random guess it'd be that it's much slower due to
more overhead in the busses, the OS.

I'm kinda lost here as if I type the ping request from the router box
itself, how is that any different than if I physically moved the DSL modem
to the other box? When I'm typing the ping request from the router box,
there is no "router" involved as the ping is comming from the box connected
to the modem so doesn't have to be masc'ed. That ping is going out through
the "hostile" nic card connected to the internet directly. If you notice
the ping requests, the one from the router has the assigned IP from the ISP
attached while the others don't.

Are you saying the speed of the computer that is sending the ping is going
to effect the ping time? Like a ping will come back faster on a P4 2.4 than
on a p1 233? I can't follow that but maybe it's true?

Since I get the same responce times from the router itself (minus maybe
..5ms) and from the network it doesn't seem to "take a significant amount of
time" to forward the packets to me. The only overhead the box has is the
network card drivers, the linux kernel and the few services that IP masc
and named takes up. It's all run from the commamd line and auto set's up on
boot through scripts. I'd think a PCI buss throughput is much higher than a
10/100 network much less a DSL connection.

 

[kony]

I'm kinda lost here as if I type the ping request from the router box
itself, how is that any different than if I physically moved the DSL modem
to the other box?

I did not know your network configuration and so just suggested to
keep the test as simple as possible.

Are you saying the speed of the computer that is sending the ping is going
to effect the ping time? Like a ping will come back faster on a P4 2.4 than
on a p1 233? I can't follow that but maybe it's true?

I don't know. I had thought that it would be since the faster system
should be able to execute the command and interpret the result faster,
that the only thing the slower system would do equally fast was at the
network card, network. Maybe issusing the command isn't what starts
the timer?

Since I get the same responce times from the router itself (minus maybe
.5ms) and from the network it doesn't seem to "take a significant amount of
time" to forward the packets to me. The only overhead the box has is the
network card drivers, the linux kernel and the few services that IP masc
and named takes up. It's all run from the commamd line and auto set's up on
boot through scripts. I'd think a PCI buss throughput is much higher than a
10/100 network much less a DSL connection.

That's a good point, but then again this minimal overheat, is bound to
be a little additional overhead not seen by the hardware router.
Maybe it is .5ms, but if so then I'm doubting that there's much
difference at all between a hardware and software router. At any rate
I think I was wrong about my router adding 5-9ms too, so maybe I
should just "shutup" and let a real networking guru step in ;-).


Dave

 

[Stacey]

kony wrote:

That's a good point, but then again this minimal overheat, is bound to
be a little additional overhead not seen by the hardware router.

A hardware router still has to masc the IP address from the internal
network, answer pings (or block them), forward packets to the right machine
etc. Maybe there is no "os" (or is there an embeded OS of sorts?) in a
hadrware router but a linux kernel isn't much more than what loads on a dos
boot floopy so I can't imagine it's that much overhead. Maybe if someone
was typing to use XP as a router on an old maching it could be bad news..
And given most hardware routers have nowhere NEAR the procesing power that
even a P1 233 has, I can't imagine the hardware routers are faster.

Do you know what sort of "processor" these use to do the masc, router and
firewall rules etc? Just curious now myself what they use to do their work.
Given they don't have a fan, it can't be much!

 

[kony]

kony wrote:



A hardware router still has to masc the IP address from the internal
network, answer pings (or block them), forward packets to the right machine
etc. Maybe there is no "os" (or is there an embeded OS of sorts?) in a
hadrware router but a linux kernel isn't much more than what loads on a dos
boot floopy so I can't imagine it's that much overhead. Maybe if someone
was typing to use XP as a router on an old maching it could be bad news..
And given most hardware routers have nowhere NEAR the procesing power that
even a P1 233 has, I can't imagine the hardware routers are faster.

Do you know what sort of "processor" these use to do the masc, router and
firewall rules etc? Just curious now myself what they use to do their work.
Given they don't have a fan, it can't be much!

I have a new model of (something-or-other) router "somewhere" around
here, if I can find it I'll crack it open and see what's inside.... at
any rate it was a low-end unit, probably typical of consumer gear.

I don't know about all routers but many use RISC processors running at
around 100MHz (possibly even faster now), perhaps quite specialized
for this application, mimimal external parts needed. That probably
outperforms a P233 easily BUT it could be that other parts of the
router, bus and memory, are crippled to minimize cost on a
consumer-oriented product.


Dave

 

[kony]

A hardware router still has to masc the IP address from the internal
network, answer pings (or block them), forward packets to the right machine
etc. Maybe there is no "os" (or is there an embeded OS of sorts?) in a
hadrware router but a linux kernel isn't much more than what loads on a dos
boot floopy so I can't imagine it's that much overhead. Maybe if someone
was typing to use XP as a router on an old maching it could be bad news..
And given most hardware routers have nowhere NEAR the procesing power that
even a P1 233 has, I can't imagine the hardware routers are faster.

Do you know what sort of "processor" these use to do the masc, router and
firewall rules etc? Just curious now myself what they use to do their work.
Given they don't have a fan, it can't be much!

I found that router I had, it's a Belkin F5D5231-4 (basic budget
router with 4 port switch). Inside is a Conexant 84200 (ADMtek
ADM5106) processor, a ARM based processor @ 75MHz, 8MB of "PC150"
SDRAM (6ns). It appears to be very similar to a D-Link 604 rev 1,
perhaps the 704 series also.

I'm not sure of the OS but the D-Link 604 seems to be running a
version of ThreadX - http://www.expresslogic.com/txtech.html
I'm not at all familiar with ThreadX but the linked page lists it's
minimum footprint as 2.5KB (yes, kilobytes).

Now that I know this, I still have no idea how it compares to a
Pentium box running 'nix. Oh well.


Dave

 

[Stacey]

Now that I know this, I still have no idea how it compares to a
Pentium box running 'nix. Oh well.

Thanx for the info..