NJ_Tech
asked on
Cannot get gigabit speed
I'm in the process of upgrading my home LAN to gigabit speed but so far I'm having no luck whatsoever going beyond 100.
I just bought two Netgear switches--a GS105 for my home office and a JF516 for the basement. Any device with a gigabit NIC plugged directly into the either switch results in a gigabit connection, according to the idiot lights, but if I connect the two switches together over about 75' of Cat 5e, the ports on each switch connected to that link shows only a 100MB link. I've tried it on two different runs of 5e going between the basement switch and office, too, with the same negative result.
Is there some trick I'm missing? I'm using 5e for the long runs (in the walls) and cat 6 patch cables. If there was crosstalk in the longer runs, would that be an issue? I don't mind somewhat degraded performance (if it ran at 400 or 500 mbps I wouldn't complain), but I can't even show a GB link signal.
I just bought two Netgear switches--a GS105 for my home office and a JF516 for the basement. Any device with a gigabit NIC plugged directly into the either switch results in a gigabit connection, according to the idiot lights, but if I connect the two switches together over about 75' of Cat 5e, the ports on each switch connected to that link shows only a 100MB link. I've tried it on two different runs of 5e going between the basement switch and office, too, with the same negative result.
Is there some trick I'm missing? I'm using 5e for the long runs (in the walls) and cat 6 patch cables. If there was crosstalk in the longer runs, would that be an issue? I don't mind somewhat degraded performance (if it ran at 400 or 500 mbps I wouldn't complain), but I can't even show a GB link signal.
If your cat5e isn't good quality, was terminated with too much stripped on the ends or was run parallel to some power cables, you can have signal loss. A nice free software tester is Ixia QCheck http://www.ixchariot.com/products/datasheets/qcheck.html. Cat5e (not cat5) should be able to handle gigabit - I use it myself at home, with runs from the basement to the attic to the second floor.
ASKER
OK, I pushed through about 500 MB to my server and also to my NAS box and I'm getting about 290 mpbs reliably. I don't know if this is bargain-basement cat 5e or premium, to be honest. Just a box of 1000' of plenum-grade 5e that a contractor gave me 3 years ago.
290 megabit/sec for data is a realistic throughput for data over gigabit when overhead for packets are taken into account - I get around 300 myself. If you got that measure over the entire run, you have a normal network connection and it's just your lights that are not working.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Actually, I'm not sure whether it's the wiring job or not, to be honest. Up until I swapped out the switches yesterday I was perfectly happy with the 100 mbps speeds I was getting, moving average-sized files between boxen and onto my server. However, once I picked up the ReadyNAS Duo, which of course is quite the bandwidth hog, I wanted to remove one more bottleneck within my system, hence the Gig switch.
Now, I'm trying something else. My NAS box and the old file server are both wired to the 16 port switch by 3 foot commercially-made Cat6 cables and the link LEDs are showing gigabit links running. The two boxen are separated by a couple of feet. According to the speed measurements (I use DU Meter, which I'm sure is close enough to accurate for what I'm doing), I'm only getting about 95 mpbs through on peaks and 66 average, using a 36 gig disk image file as the test material.
Re-running new cable to my home office is doable but at an extreme level of difficulty and expense, and I'm wondering if it's warranted. With bone-stock NICs, 3 year old 5e cable and home-made custom-length patch cables run between the punchdown and the switch, what kinds of speeds should I realistically be getting in this very real-world setting?
Now, I'm trying something else. My NAS box and the old file server are both wired to the 16 port switch by 3 foot commercially-made Cat6 cables and the link LEDs are showing gigabit links running. The two boxen are separated by a couple of feet. According to the speed measurements (I use DU Meter, which I'm sure is close enough to accurate for what I'm doing), I'm only getting about 95 mpbs through on peaks and 66 average, using a 36 gig disk image file as the test material.
Re-running new cable to my home office is doable but at an extreme level of difficulty and expense, and I'm wondering if it's warranted. With bone-stock NICs, 3 year old 5e cable and home-made custom-length patch cables run between the punchdown and the switch, what kinds of speeds should I realistically be getting in this very real-world setting?
ASKER
Add to the above comment, once I enabled jumbo frames on the ReadyNAS, the throughput increased slightly, to an average of 115 mbps and a maximum of about 140, and that's going in either direction -- server to NAS or vice-versa.
>With bone-stock NICs, 3 year old 5e cable and home-made custom-length patch cables run between the punchdown and the switch, what kinds of speeds should I realistically be getting in this very real-world setting?
That should be able to achieve gigabit transfer rates. The fact that you actually went above 100Mbit/sec seems to indicate it is not a cable or network problem. As mentioned before, if the disks on either end can't handle higher sustained transfer speeds, it won't matter if you use gigabit. Have you tested both sides for this? HDTune is a good benchmarking tool.
That should be able to achieve gigabit transfer rates. The fact that you actually went above 100Mbit/sec seems to indicate it is not a cable or network problem. As mentioned before, if the disks on either end can't handle higher sustained transfer speeds, it won't matter if you use gigabit. Have you tested both sides for this? HDTune is a good benchmarking tool.
Do a throughput test with FTP or TFTP or some well known protocol and then figure out how long it takes to transfer x amount of data and figure out your througput.