![]() |
Networking gurus: riddle me this speed loss please
I have a vexing internet speed loss in the house...
I run a Verizon FIOS actiontec modem in the garage, theoretically spitting out 20 Mbps download / 5 upload. Yet I get about 5/1 in my home office, wired... The garage modem has 4 ports with cat5 ethernet cables plugged into it that go throughout my house. I checked speeds at the end of each cable, 3 of the 4 give me the same 24 Mbps, which is better than advertised (some are 30-50ft cables). Then the 4th is 100Ft cable running in the ceiling. At the end of that I have 12 Mbps only !!! 50% loss... Huh ??? Didn't think length affected throughput this much ! Worse, that one 100 ft cable (going to the home office) goes into a hub and splits into 3 computers and one vonage device - never all used at the same time... Out of the hub, I get 5 Mbps.. Again, WTF ? I disconnected all devices out of the hub, no matter, it maxes out at 5... out of the bad 12 to begin with ! To add insult to injury, I added an N wireless router off the primary in the garage, verified 24 mbps at the cable, yet the wireless signal spits out 3 Mbps in speedtests, in N -only mode !! I decided to enable the old b/g of the primary router in the garage again to compare and that is faster, at 6-7.. Still slow, but faster... Is it possible that I got a bad hub, a bad wireless router, and a bad cable all at once ? Or am I missing a magical setting I'm unaware of ? I'm borrowing a cable tester to test the 100Ft cable wiring... Would passing in the ceiling close to electrical wires or neon (despite power being turned OFF) explain a 50% speed loss ? |
Cat 5 is good for 100 meters so you are well under that.
Always best to stay at least 1 foot away from electrical runs. If you have to cross one do it as close to perpendicular as possible. Are you using shielded or unshielded cable? While shielded may intuitively sound better it requires more care/extra work to install properly. If done wrong it will cause problems. Don't know if you can get into the ceiling to inspect thee cable but sometimes they can get kinked while pulling the run, does not help things. Remember that in mixed device settings you will get the speed of the slowest device. |
OK.... I am guessing the upstairs hub is old and I will get a new switch anyway...
The 2 things that trouble me is that I get 24 at the modem and 12 at the end of the long cable... exactly half. I'm not a network guy, but I remember something about duplex.... It was laid out nicely before the ceiling was closed, no kinks... I'll test it tonight. If it's the cable it's going to be a real pain to thread a new one ! The other annoying thing is to see 24 Mbps at another cable's end, plug that into an 802.11 N and observe 3 Mpbs via wireless while standing in front of the router... Wow ! |
Also, when it comes to "N" wireless, there were a ton of devices that were released before the specification was ever standardized and finalized. This means that there can be some issues when mixing brands, so if your wireless router is of a different manufacturer than the card in your laptop/device, it can result in less than theoretical speeds when the negotiations between devices are done.
|
Could be a bad cable causing the negotiated connection to drop to a lower speed or duplex. Or... the upstairs hub is old as you say and negotiating down to a lower rate or duplex for you. If your making your own cable runs get a tester, they are about $20 for a decent one that will show faults and loss.
|
Quote:
|
Hi - I would reduce the number of variables that could be causing the slowdown. I would attack the slow wired connection first.
1. Swap the ports on the Actiontec and check speeds. 2. Remove the hub and connect the 100ft cable directly to a PC and check speed. If still slow, not the hub but you should still replace it with a switch. 3. Move a PC to the Actiontec and connect and check speed. If still slow, not the cable As for the wireless issue - depends on the firmware on your N router. Also the wireless settings on both your N router and the Actiontec. I have a N router connected my Verizon router (off a switch) w/ wireless disabled and DHCP enabled. Good luck |
Some of the Cat 5 cables will only give you 100Mbps ethernet. I had to change all of the cables at the office here to true Cat 6 so we are a 1GB network.
|
Switches are good, hubs are bad. If you're using a hub, you may as well just twist all of the wires together. A hub will not be full duplex, and will create collision domains which will make things much slower.
Try the other wireless flavors for different speeds. Find the most up to date drivers for your wireless adapter. |
Quote:
All should be set to Auto but it is possible to set them individually for specific 10 and 100 half or full duplex |
How is youre cable terminated? My guess the cable is probably fine but the end or ends are causing th issue. Technically that cable should be terminated into a wall plate and a patch cable to your switch (hopefully a switch and not a hub).
You status on the NIC on the computer will show your status as well, including half/full duplex when you are connected to the cable. |
Quote:
Also - if your 100 ft cat-5 cable is kinked, it is junk. I would suggest running another cable from the modem to upstairs as a test. Just run it through the house if you have a long enough cable, and see how it fares. -Z-man. |
Yes on all of the above. Thanks... I will move the long cable to another port first, and retest. I have a new 100Ft cable which I will also test (still coiled) on location before I even try to snake it there ;-) I have a switch on order - that part I suspected so I'm not worried.
I'm still slightly baffled by the long-cable-only 50% loss (total loss from a kink I would understand, but 50%?) - that's before the hub - and by the N router's crappy output vs a b/g, but I will re-update firmware and see.... Thanks guys, good enough for starters... I spent 1 hour last night spelling my name to tech support in India before they'd reset on their end, then an hour running around the house with a laptop running speedtests at the end of every wire ;-) By God I'll get my 20/5 ! |
It isn't a hose, kinks don't matter on a run like that.
What are you doing that 12mb down isn't good enough and 24mb will fix? If you want to know your network throughput you should test things locally between computers on your LAN, not all the way to your ISP. |
Quote:
So there's something wrong, and that bothers me. I probably don't *need* 24 download, but I sure could use the full 5Mbps upload at times... |
Are you getting bits and bytes confused? 3MBPS is actually pretty fast for wireless, if you ask me.
|
Quote:
Open a web browser and put in 192.168.1.1, that will take you to the routers admin login panel. User: admin Pass is: password or password1 If those passwords don't work then reset the router, on the back use a ball point pen and push the reset button for 10 or 20 seconds. Make sure to assign a different password unique to you. |
+1 on not running the network cables near electrical wires.
Network cables really struggle if they are run over fluorescent light fixtures. These are VERY noisy. So are electric motors. |
Quote:
Rule of thumb for Cat 5 is that the cable can be bent to any radius larger than 4 times it's diameter. |
You have a lot of stuff going on there that could be causing your slow speeds.
First and foremost are you testing the connectivity at the ends with the same computer? If not you could simply be seeing the end results on slow computers are different than faster ones. After that - the hubs. Hubs are repeaters. They basically allow you to plug a bunch of computers into a single wire. All of those computers/devices share that one wire's bandwidth AND hear all that talk and if it is a FULL DUPLEX hub the other devices will only talk when nobody else is talking (otherwise that collision light will go off a lot). Get rid of that hub and in its place put a switch. The difference is that the switch will not repeat everything on all ports. The Switch will keep each device's traffic separate upstream. Downstream of course it's still that long 100mb connection. HAve you tested BEFORE the hub at the end of that long cable? I think you said you did - at that point I would question the integrity of that cable. On a real router I would be looking for errors on the interface but on these home use devices you usually can't see that. Consider replacing the long cable - I see you're already doing that - test it first. Regarding the 'N' wireless device - there you may simply have a quality issue. Something can be labeled 'N' and support the standard easily but not actually be able to do it. What are your expectations on that wireless? I wouldn't expect more than 20mbps down in the best of conditions from a home use device. The best APs I am familiar with (though I'm biased of course) are the Cisco 3500 series which I have personally tested at about 70mbps up/down. I'm testing the 3600s now and am hoping for better. I would first concentrate on the wired speeds and get that worked out and then focus on the wireless issues. Do I have a home use device that I recommend? Not really - it seems to be a 'crapshoot' on that front though I do like the Linksys gear (again, biased...). |
All times are GMT -8. The time now is 10:25 PM. |
Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2025, vBulletin Solutions, Inc.
Search Engine Optimization by vBSEO 3.6.0
Copyright 2025 Pelican Parts, LLC - Posts may be archived for display on the Pelican Parts Website