Linksys Router Speed

[Gun Geek]

Picked up a Linksys A+G router/WAP yseterday.

Hooked it up to my existing operational cable MODEM, no problems getting it to work. However when I visit sites, performance is significantly slower than when directly connected to the MODEM. This is via hardwire from the pc to the router - haven't even gotten to the wireless stuff yet.

When I go to a speed test site (2wire.com, dslreports), the speed is about 1/10 through the router (125 Kbps) as it is via the direct connect (1200 - 1300 kbps).

I expected som performance hit, 10% - 20%, but not this!!!

Spent a couple of hours on the phone with Linsys tech support. They're stumped.

Anybody got any ideas?

[George]

First thing to try would be to set up your computer to use a static IP instead of DHCP (Auto). That usually adds speed to any Ethernet LAN.

It is also possible that you have a 10 base T Ethernet card instead of a 10/100 card in the computer. If so, that and DHCP being enabled for the computer could be robbing you of speed.

Also check to see if the buffer settings in the computer are set to their max. Low settings there in the LAN control panels can also rob you of Ethernet speed.

--

Regards,

[Gun Geek]

First thing to try would be to set up your computer to use a static IP instead of DHCP (Auto). That usually adds speed to any Ethernet LAN.

Linksys had me try that - no change

It is also possible that you have a 10 base T Ethernet card instead of a 10/100 card in the computer. If so, that and DHCP being enabled for the computer could be robbing you of speed.

The modem is a only 10. The Router and PC are 10/100. The MODEM/PC connection (at 10) is the fast configuration. Could there be a problem or config issue with the 100 function on the PC (sure there could) how could I tell?

Also check to see if the buffer settings in the computer are set to their max. Low settings there in the LAN control panels can also rob you of Ethernet speed.

More detail on this, please?

[George]

The I/O buffer settings are in the LAN control panel settings under the advanced section (I think). I don’t work in Windows all the time so I am a little rusty memorywise as to exactly where they are, but they are there in the TCP-IP area somewhere.

Basically, these settings set the packet size that is fed to and accepted from the NIC. If these are set too low, it can cause a large speed drop. They are usually already set properly for the Ethernet card in the machine, so do remember the setting there before you change it because if you get any trouble you will need to restore it. Larger is usually faster if the NIC can handle it.

--

Regards,

[Paul B]

Maybe the problem is not the equipment but the test. I ran my DSL on the 2wire test and it said 196kbps, but I went to two other speed test sights and they indicate 1300kbps. I certainly have no speed problems downloading big files. ????????

[George]

Good point about the speed test being possibly suspect. If you have multiple machines running on the LAN, try a transfer of a known size file between the machines and time it. That will give you a rough idea of the LAN throughput speed.

Then go to the web and try downloading a 3-4 MB file and time it. If it comes down in 15 seconds or less, then you have the speed and the test is wonky. If it takes more than a minute, then the modem/router connection is wonky (or the site is jammed with traffic possibly).

--

Regards,

[Yardbird]

I just tried 2wire and both times it gave me numbers between 155 and 180 kbps, but when I pushed the BACK button it appeared to run the test again and gave me 255000 kbps. Twice! Strange!

[Gun Geek]

I'm not religious about what 2wire tells me (I use oters as well). It is only part of the test.

Loading a "rich" page (like Yahoo) is noticably slower through the router than directly connected to the MODEM.

If I'm really froggy, I'll time a download. The absolute numbers are not really that relaible, but the magnitude of the difference and the noticbly longer load times are part of the decision.

[Rob Boudrie]

I just tried 2wire and both times it gave me numbers between 155 and 180 kbps, but when I pushed the BACK button it appeared to run the test again and gave me 255000 kbps. Twice! Strange!

Looks like the test found the test file in the browser cache and used an elapsed time which did not include actual data transfer in the calculation.

[George]

Rob makes a good point about clearing browser cache before running these tests a second time.

Nothin’ like local cache for speed ;-)

--

Regards,

[geezer-lock]

This assumes WinXP.

Go to Control Panel>Network Connections then right click Local Area Connection. On the General tab you will see Connect using: which will indicate your NIC. Click the Configure button to open the Connection Properties dialog box. Click the Advanced tab, then in the Property box choose Link Speed & Duplex. The Value window will display your current setting. If Auto Detect is selected, change to 10Mbps/Half Duplex. Click OK to save the settings.

Check your performance (remember to clear the cache). If that works you are done, if not try the connection on full duplex.

geezer

[mg1199]

Another idea you might try is to see if you can place your computers IP address into the DMZ zone of the router. This will put your computer outside of your firewall and bypass the normal router functions. test the speed after you have done that. (choose one of the speed sites, and just stick with that. speed can vary from site to site becuase of hte number of hops that your packet will have to travel through. if there are only a few number of hops, and you aren't losing a lot of ms because of it, then it will be more realistic. you can find the # of hops your packet travels to that site by opening up a dos window and typing :> ping xxx.xxx.xxx.xxx which is the ip address of the speed site that you are using. this will give you the relative "internet distance" that your info (packet) has to travel.)

afterwards disable the DMZ function, which will put your computer behind the firewall again, and then re-test the speed.

if there is a significant difference in speed, your router could have one of the filtering features on (although it shouldn't be that drastic of a speed difference). what you should do is take a look at all the router settings, and turn off any sort of filtering, and then repeat the test. if you can do a complete factory reset of the router, that would be better. usually there is a reset button somewhere on the back of the router than you can press while you power it up, which should put all the settings back to factory.

if there isn't a speed difference after you have tried the DMZ, well then.... i'm not exactly sure. if you have tried the same internet speed test site with both the router connected, as well as not connected (on the same computer also), then i would think that there is something defective with the router. still have the receipt?

[Rob Boudrie]

Another thing to try is to turn DHCP off in the router, connect a router of known speed to the cable mode, and connect one of the ports on your wireless router to a port on the known good router (use a crossover cable for this, although some routers have autosensing ports so a straight cable may work as well). You will then be using the router as a switch with wireless capability, but not using the NAT or DHCP features of the wireless unit - see if that makes a difference. If the performance slowdown is in the network address translation feature of your router, this could help.

I use this topology at home, but not for performance reasons. My router w/NAT is in the basement connected to a second switch and all my systems connect at this point. I placed a wireless unit upstairs so that I would have best signal coverage, and use the DHCP from my main router, using the wireless unit as a switch w/wireless, but not as a router, NAt bor or DHCP server. Works great.

The only problem is askin customer support about this. Most of the consumer level customer support types are trained to support only the most simple basic configuration.

Rob

[raz-0]

are you using the A part of it or the G part of it? And do you have a cordless phone?

If you have A and a 5Ghz phone, they can interfere. Interference = lower speed. If you are using G and you have a 2.4GHz phone, they can interfere. If you have a panasonic 5Ghz phone, it'll interfere with both depending on the model of phone as panasonic kind of lies on some of it's models and uses both 5ghz and 2.4Ghz. If it is G, the microwave can also interfere with it when running (as well as any 2.4Ghz cordless).

Also you might want to DL netstumbler for you PC and see who else is running wireless in the neighborhood (if applicable), and make sure you are on a different channel than they are to avoid interference.

Also make sure you are connected to YOUR router and not someone else's. Turn on all your security to make sure nobody else is connected to YOUR router as sharing the signal can slow you down very much.

[Gun Geek]

are you using the A part of it or the G part of it? And do you have a cordless phone?

If you have A and a 5Ghz phone, they can interfere. Interference  = lower speed. If you are using G and you have a 2.4GHz phone, they can interfere. If you have a panasonic 5Ghz phone, it'll interfere with both depending on the model of phone as panasonic kind of lies on some of it's models and uses both 5ghz and 2.4Ghz.  If it is  G, the microwave can also interfere with it when running (as well as any 2.4Ghz cordless).

Also you might want to DL netstumbler for you PC and see who else is running wireless in the neighborhood (if applicable), and make sure you are on a different channel than they are to avoid interference.

Also make sure you are connected to YOUR router and not someone else's. Turn on all your security to make sure nobody else is connected to YOUR router as sharing the signal can slow you down very much.

<{POST_SNAPBACK}>

Haven't even messed with the wireless part yet. This speed problem is via wire connection.

I will exchange the router and see what happens. Probably today.

[richardschennberg]

You might try an SMC Baricade router, such as the SMC7004VBR. One of the better network consultants in our area uses them because his clients rarely have problems once it is installed.

Richard

Schennberg.com

[Clyde]

I'm guessing you have done a firmware update to the router?

[Gun Geek]

I'm guessing you have done a firmware update to the router?

<{POST_SNAPBACK}>

1st thing linksys did after they took my and address

[OpnBlstr]

Try http://myspeed.visualware.com .

Their speed test is generally very accurate. It's the VoIP test at the top of the page.

Try setting your MTU on the router to either 1492 or 1500. Some connections like one over the other. Ask your provider what works best with their system.

Go to microsoft.com and DL a large file. If you get it at around 100K (assuming 1.5 DSL) everything is fine.

DHCP vs static IP shouldn't make any difference. If you have a duplex issue though then that will show up in a big way. Try manually setting both your NIC and the router ports to 100FD. In all actuality, unlike 5-10 years ago, modern Ethernet hardware is pretty dumbproof. If you set your NIC for AUTO it will choose the best setting to match your router interface.

Are the network cables you are using in good shape? Try new cables. A single bad wire will still work (TCP/IP is incredibly resilient!) but throughput will suck bad. You don't need the most expensive cables either. Cheapy AllenTel cables work 100%.

For security, if your Linksys is a WRT55AG and your wireless clients support it, use WPA/PSK instead of WEP. It's much more secure and changes the keys automatically at a preset interval.

FWIW, I have the WRT55AG connected right now and just got 4.22M down/530K up on the speed test so the Linksys routers CAN support the throughput.