Internet connection speeds....

[Tim Kirkwood]

I have been playing with some differant setups here at the house and compairing them using the www.smugmug.speedtest.net (http://www.smugmug.speedtest.net) test.

I like that it keeps a list of your results for you.

Anyways....

With my Wifi connection I get in the neighborhood of around 1300-1500 down and around 365 for uploads.

With a hardwire to the router I get about 4350 down and the 365 for up,

Now if I take my router out and go straight to the cable modem i get up to almost 5000 down and still the 365-368 up.

Was wondering if anyone could suggest ways to get the router to give me that 5000 down with a hardwire connection? Cant figure out why it would be such a bottle neck.

Also has anyone upgraded from the wireless G to a wireless N network? How big of a differance have you seen? i would like to keep using my wireless but with such a huge differance I am considering running a line from the router to the living room where I normally sit with the laptop.

Thanks for any tips you can provide D'grinners!

Tim

[photocat]

it is worth a try with a hard connection (ethernet only between computer and where the cables comes in)

5000 is so fast, I can never get more then 1 MB as I am at the end of a line.
My upload is not even a 100... You lucky bum!

[Tim Kirkwood]

it is worth a try with a hard connection (ethernet only between computer and where the cables comes in)

5000 is so fast, I can never get more then 1 MB as I am at the end of a line.
My upload is not even a 100... You lucky bum!

Cat,

What type of connection do you have? And where are you located and are you testing to the smugmug test in san jose, cal?


Tim

[LuckyBob]

Provided your modem is *just* a modem, pulling your router out of the network will also take out your hardware firewall. The router's a slight bottleneck because it's filtering everything for you. It's not worth the risk and management hassle (of properly securing the machine itself) to plug a computer directly into the modem.

Also has anyone upgraded from the wireless G to a wireless N network?
Won't make any difference. Wireless G is (theoretically) 54Mb/sec, and your internet connection is maxing out at 5Mb/sec. Even with WEP overhead and realistic performance, G will still sustain around 20Mb/sec, so it's not going to be a bottleneck. If you're getting horrid performance over WiFi, your network may be on an overloaded channel.

http://www.speedtest.net/result/76930824.png (http://www.speedtest.net)

[Tim Kirkwood]

Provided your modem is *just* a modem, pulling your router out of the network will also take out your hardware firewall. The router's a slight bottleneck because it's filtering everything for you. It's not worth the risk and management hassle (of properly securing the machine itself) to plug a computer directly into the modem.



Well I only took the router out for a speed check. I do not want to take it out of the loop for good but was wondering if anyone knew of any settings that are best for optimal speed?

I also am very curious to some first hand testimonies of the wireless "N". They advertise that it has increased distance and speed, just wondeirng if its worth the price to upgrade.


Tim

Edit: Also, bob, what type of connection and where are you. 10mb is nice!

[photocat]

I am in jolly good England... People get 8 MB in London, but not uphere...
We are lucky, at least we have broadband. Well, what we call HERE broadband... grin
So no, I am not doing the San Jose test...

[Tim Kirkwood]

I am in jolly good England... People get 8 MB in London, but not uphere...
We are lucky, at least we have broadband. Well, what we call HERE broadband... grin
So no, I am not doing the San Jose test...

I can upgrade my service to a 8mb here but not sure how much more it cost. Try the san jose test out of curiosity lol


Tim

[Fred Maurer]

Thanks for the link, I tried it with a Belkin pre-n notebook wireless card and Comcast cable and got 12,563 kb/s download, and 364 kb/s upload. Using direct wire laptop to router = no significent change

Different laptop with an intel 2915 wireless network card, with the same router et al = 308 kb/s down and 137 kb/s up. Using the second laptop with direct wire to router = 12202 kb/s down and 369 kb/s up

The card obviously makes a huge difference. No wonder I don't have the patience to use the 2nd laptop!

[Tim Kirkwood]

Thanks for the link, I tried it with a Belkin pre-n notebook wireless card and Comcast cable and got 12,563 kb/s download, and 364 kb/s upload. Using direct wire laptop to router = no significent change

Different laptop with an intel 2915 wireless network card, with the same router et al = 308 kb/s down and 137 kb/s up. Using the second laptop with direct wire to router = 12202 kb/s down and 369 kb/s up

The card obviously makes a huge difference. No wonder I don't have the patience to use the 2nd laptop!

Thats some nice speed Fred, is that your home connection? Who is your ISP?

[Fred Maurer]

Thats some nice speed Fred, is that your home connection? Who is your ISP?
Comcast Cable The pre-n card wireless card is just about equal to being hardwired and I can go down the road about 40 yards from the house before losing the connection.

[LuckyBob]

I also am very curious to some first hand testimonies of the wireless "N". They advertise that it has increased distance and speed, just wondeirng if its worth the price to upgrade.

I can vouch for the increased range and speed of Pre-N equipment (due to better noise filtration and stronger antennas), but for surfing the internet, the bottleneck will be the 'net connection itself. Even 802.11b networks are plenty fast for surfing as long as your internet is >2-3Mb/sec. If range is an issue, ugrading antennas is a *much* cheaper route to stronger signal strength. Check out HyperLink's (http://www.hyperlinktech.com/web/antennas_2400.php) selection of 2.4GHz antennas. We've built wireless networks based on 802.11g with their antennas spanning five city blocks line-of-sight.

I'm in Seattle using Comcast cable, and if I test to local servers I get well over 15Mb with that test. The SmugMug server test seems to be on a bit of a slower backbone - I've found other servers (including our own rack in Peer1) which can burst to 25Mb/sec. Note that Comcast's speed test results tend to be quite skewed since they place no bandwidth restrictions on the modem for the first few seconds that a connection is established; *if* they trigger the bandwidth cap, it'll throttle back down to 6Mb/sec. I will say they're quite sloppy at doing so, though; fairly often I can finish 100MB+ downloads without being capped - see the atached image below - 2477.9KB/sec = 19.8Mb/sec.

As far as router settings, there's usually not much you can do internally in the router itself to change its performance aside from messing with the MTU, although the default of 1500 (in most routers) is technically optimal. A better router would help, though. Personally, I'm running a Linksys WRT54GL with firemware v4.71.1, Hyperwrt 2.1b1 + Thibor15c, which is MUCH better than the stock Linksys firmware.

[LuckyBob]

Different laptop with an intel 2915 wireless network card, with the same router et al = 308 kb/s down and 137 kb/s up.[...]

The card obviously makes a huge difference. No wonder I don't have the patience to use the 2nd laptop!

There's something seriously wrong with that card if it's operating that slowly :huh

[ian408]

I am in jolly good England... People get 8 MB in London, but not uphere...
We are lucky, at least we have broadband. Well, what we call HERE broadband... grin
So no, I am not doing the San Jose test...<table border="1" bordercolor="black" width="505" bgcolor="">
<tr>
<td> Location </td>
<td> Upload </td>
<td> Download </td>
<td> Latency </td>
</tr>
<tr>
<td> San Jose / Kariachi</td>
<td> 96 </td>
<td> 316 </td>
<td> 335 </td>
</tr>
<tr>
<td>San Jose/London </td>
<td> 316 </td>
<td> 701</td>
<td> 185 </td>
</tr>
<tr>
<td> San Jose/San Jose </td>
<td> 326</td>
<td> 1287 </td>
<td> 66 </td>
</tr>
<tr>
<td> San Jose/Guatemala City </td>
<td> 304 </td>
<td>1286 </td>
<td> 112 </td>
</tr>
</table>

(*) Done via wireless laptop to Linksys Wireless G router and
AT&T DSL connection.


Generally, latency is the big killer in the overall picture and many
folks don't realize that. I'd also

If it were me, I'd be looking for a new router.

[Tim Kirkwood]

Check out HyperLink's (http://www.hyperlinktech.com/web/antennas_2400.php) selection of 2.4GHz antennas. We've built wireless networks based on 802.11g with their antennas spanning five city blocks line-of-sight.

Thanks for the cool link! Would be cool to have my network available all over the neighborhood like that!

24db! ---> http://www.hyperlinktech.com/web/hg2424g.php

I always worry about replacing antennas on things. I worry that I am going to screw up the load balance and cause something to burn out. How do I know which antennas are safe for my router?

Tim

[Tim Kirkwood]

Personally, I'm running a Linksys WRT54GL with firemware v4.71.1, Hyperwrt 2.1b1 + Thibor15c, which is MUCH better than the stock Linksys firmware.

I have a linksys router, how do I find this other firmware that you speak of? What are the advantages to it over linksys's firmware?

Tim

[LuckyBob]

I always worry about replacing antennas on things. I worry that I am going to screw up the load balance and cause something to burn out. How do I know which antennas are safe for my router?

Tim

All of 'em are :D. I've never once had a problem replacing antennas with stronger ones. Regarding the firmware, a handful of Linksys products are based on a Linux kernel (which means they're open source), and there's a handful of third party updates to the firmware which VASTLY improve features/stability, plus you can SSH into your $40 router, which is just cool :thumb! Here's a bit of info: http://en.wikipedia.org/wiki/WRT54G#Hardware_revisions

[Tim Kirkwood]

All of 'em are :D. I've never once had a problem replacing antennas with stronger ones. Regarding the firmware, a handful of Linksys products are based on a Linux kernel (which means they're open source), and there's a handful of third party updates to the firmware which VASTLY improve features/stability, plus you can SSH into your $40 router, which is just cool :thumb! Here's a bit of info: http://en.wikipedia.org/wiki/WRT54G#Hardware_revisions

Thanks,

I found this statement(below) on the Wiki and was wondering of those antennas on that site are reversed polarity?

" RP-TNC connectors are widely used by Wi-Fi (http://en.wikipedia.org/wiki/Wi-Fi) equipment manufacturers to comply with specific local regulations i.e. FCC (http://en.wikipedia.org/wiki/Federal_Communications_Commission) which are designed to prevent consumers from connecting aerials which exhibit gain and therefore breach compliance"


Tim

[LuckyBob]

I found this statement(below) on the Wiki and was wondering of those antennas on that site are reversed polarity? [...]

Tim

Depends on the antenna - most of them use N-Male connectors by default since it's the beefiest connector, but a lot of the antennas are available with different connectors, including RP-TNC and RP-SMA. There's a cool feature on their site when viewing some product details; when you pick a particular connection type, click on "Radios which use <connector type> connectors", and you'll get a fairly comprehensive list of the equipment that uses said connector. Linksys is pretty devoted to RP-TNC

[Tim Kirkwood]

Just an update to this thread, Roadrunner has made alot of speed increasing changes latley. Here is a new speed test with my wireless router in the mix and in the middle of a SAturday afternoon.


Much better results.
http://www.speedtest.net/result/161829572.png (http://www.speedtest.net)

[Fred Maurer]

Since you bumped this back up, thought I'd do an update too. New router and modem due to lightning strike. Belkin N-1 wireless router now, same
ISP-Comcast
http://www.speedtest.net/result/161881139.png (http://www.speedtest.net)

[David_S85]

Is this bad?
http://www.speedtest.net/result/161890466.png (http://www.speedtest.net)

what? :dunno

[DavidTO]

Is this bad?
http://www.speedtest.net/result/161890466.png (http://www.speedtest.net)

what? :dunno


No fair, using dial-up to get a sympathy vote!

[David_S85]

No fair, using dial-up to get a sympathy vote!

But.... It's my only line!
(old joke - crowd laughs)

[Tim Kirkwood]

Since you bumped this back up, thought I'd do an update too. New router and modem due to lightning strike. Belkin N-1 wireless router now, same
ISP-Comcast
http://www.speedtest.net/result/161881139.png (http://www.speedtest.net)


Thats some connection you got there Fred! I am pretty jelous!


Tim

[jdryan3]

http://www.speedtest.net/result/161910326.png

http://www.speedtest.net/result/161911340.png

[Cameron]

Smugmug used to be one of the fastest connections for me on speedtest.net. In recent weeks it has looked like this:
http://www.speedtest.net/result/161937629.png (http://www.speedtest.net)

However, using the San Francisco server (not far from you guys), I get this:
http://www.speedtest.net/result/161938061.png (http://www.speedtest.net)

:scratch
I'm in the Kansas City area, using Time Warner Cable (rated 7mbps)

[Eric&Susan]

I'm using Comcast and have my modem hooked the the wirless router which is then hooked to my desktop via ethernet and I get:

http://www.speedtest.net/result/162319864.png (http://www.speedtest.net)

Eric

[jdryan3]

Interesting article here (http://www.pcmag.com/article2/0,2817,2335749,00.asp) from PCMag about ISP performance and satisfaction. Their point is typical speed tests live in a vacuum, and at times (often?) don't mimic wait people actually experience in their web cruising lives.

[ian408]

Interesting article here (http://www.pcmag.com/article2/0,2817,2335749,00.asp) from PCMag about ISP performance and satisfaction. Their point is typical speed tests live in a vacuum, and at times (often?) don't mimic wait people actually experience in their web cruising lives.
It's important to note that a servers physical location relative to your target system has no bearing on the speed tests. Why? Your target may be on a different network and/or the route to the two machines may be different.