Extend range of WIRED system

[Swampster]

I have a system with a RV082 Cisco router (8-port) hosting 7 stations all well within 50' and the 8th port uplinking to a LinkSys 16-port unmanaged switch.

All works fine on the systems hooked to the switch with the exception of two systems that are connected as follows: CAT6 from the switch, about 150+' to a linksys route with the DHCP server turned off and that serves as a switch to the two slow stations that are within 15 feet of the linksys router.

If I have both of the systems on, it is so slow it feels like Dial-Up. If I disconnect either, then it is noticably faster, but still no where near what the other 4 stations out of this switch or the seven stations out of the Cisco router are doing.

I really need to have both of these stations running at the same time with a decent throughput . . . any suggestions?

 

[spidey07]

How is the cable terminated? That's too long for a patch cable and crimping ends onto cable normally yields category nothing rated cable. You'll need solid core cabling punched down into keystone jacks and then patch cables on each end.

If installed correctly you will get perfect performance of up to 100 meters - 90 meters of solid core cable with 5 meter patch cables on each end.

 

[Swampster]

Hadn't thought of that . . . Thanks, I will give it a try and report back tomorrow.

 

[Swampster]

OK fellow networkers . . . status update.

I gave thought to Spidey07's suggestion, and before I made any changes there I decided to do a loop-back test to test the continuity of all 4 pairs, and they came out perfect. While I pondered my next move, it dawned on me that CAT6 has Feet Markers, just like CAT5, so I checked and discovered that it was a lot further than I had guessed.

Total length from switch to switch is 303 feet, plus about 20-25 feet for the cable that runs down under the sound booth, under the floor, and up in the vestabule where the Welcome Center computer resides.

OPPS!!! A little too close to upper limits . . . no wonder it's a little slow! In consideration of the fact that the highest traffic on this run would be if both computers boot at the same time, thus both would be checking for OS and Security updates, I decided to try getting one computer up and settled, then boot the next computer, and that is a tick slower than I would like, but it is usable.

The Sound Booth computer only needs the network a couple of times during the week to set up the Sunday morning and evening music and Sermon Notes, etc., and the Welcome Center is only on and using the network on Sunday when the Sound Booth is on, but not doing anything over the network, so I think we might be able to get by as.

UNLESS . . . someone has a suggestion on how we might mitigate the basic problem of a run that is too long.

 

[ScottMac]

Mechanical loopback doesn't tell the tale. Continuity and polarity aren't enough.
IF the cabling is not terminated according to EIA/TIA 568 (like {pair}{pair}{pair}{pair}, which is VERY WRONG), it will not perform to spec (distance and throughput efficiency).

If you have it terminated to jacks, it might be worthwhile to verify that the twist is maintained up to the punch, that there is no more than 1/2 inch (3/8" for Cat6 and above) of exposed pair, that there are no bends exceeding minimum bend radius (~1" radius usually), and the other specs.

Make sure your patch cables are also store bought and not damaged.

Good Luck

 

[xSauronx]

i think the easiest way would be a wireless bridge between the two areas. a couple routers to act as bridges
would probably work (depending on interference in the place, and the physical layout of the building). one AP and an adapter in each may work, but if the machines are close enough together to use patch cables i think a bridge that they can be plugged would be better

 

[Swampster]

ScottMac,

Thanks for the reply!

To the best of my ability, the RJ45 termination is to specs, and the connection from the Router to the base switch to the remote switch to the further of the two computers is reasonably useable as long as only one of the computers are actually doing something over the network at a time.

The problem is two-fold. #1, the unsophisticated (spell that volunteer) users are likely to get anxious if the network response isn't like they are used to it on their home Internet connections. And, #2, you and I both know that as soon as we get them used to these added workstations, they will want to expand on their usage, and it is not likely to handle that.

I was giving consideration to a CopperLink Copper Ethernet Extender Model 2168A/EUI-2PK kit that I saw on patton.com. What are your thoughts on that type of a solution?

 

[Swampster]

xSauronx,

Thanks for the reply.

Our first idea was to get as close as I could in building one and use a 300N wireless router with the DHCP Server disabled, and an external 300N USB at the Welcome Center computer. This worked, but was not as reliable as they had hoped once you got a bunch of people in the building on a Sunday morning and doors opening andclosing in the line of sight, etc.

That is why, when they decided they had a great need to install an alarm system that they had the alarm guy run some CAT6 over with it.

 

[ScottMac]

Originally posted by: Swampster
ScottMac,

Thanks for the reply!

To the best of my ability, the RJ45 termination is to specs, and the connection from the Router to the base switch to the remote switch to the further of the two computers is reasonably useable as long as only one of the computers are actually doing something over the network at a time.

The problem is two-fold. #1, the unsophisticated (spell that volunteer) users are likely to get anxious if the network response isn't like they are used to it on their home Internet connections. And, #2, you and I both know that as soon as we get them used to these added workstations, they will want to expand on their usage, and it is not likely to handle that.

I was giving consideration to a CopperLink Copper Ethernet Extender Model 2168A/EUI-2PK kit that I saw on patton.com. What are your thoughts on that type of a solution?

300 Feet is well within normal distance for UTP, if it's terminated properly. You can run as much traffic on that as you could within 50 feet ... no operational distinction at all out to 100 meters (328 feet).

That's why it's important to follow the specs. If it's slow, that would be good motivation to investigate and ensure that all of the cabling, components, and termination are in-spec.

Good Luck

 

[skyking]

What Scott said. If it is under 300 feet then something is wrong, the distance has nothing to do with it. I've certified many drops out to 250+ feet with no latency or bandwidth issues ever. It works as well as sitting in the switch closet.

 

[Emulex]

poopy switch. higher end switches tend to manage fringe distance better. procurve is bottom of the barrel as far as decent switches go.

may not do you any good to use cat6/6a cable if you use cat5 crimps or wall jacks. or anything else in between.

you can't manage to find somewhere in between the two points and add another switch? small gig-e 5 port cheapie? that would reduce your run well within usable distance.

 

[westom]

Originally posted by: xSauronxTo the best of my ability, the RJ45 termination is to specs,

I still do not see where you confirmed each wire is only matched with its twisted cousin. Your symptoms are typical of wires that are not matched with their twists mate. You must literally get an eyelet (or some magnifying device) and confirm the colors inside each RG45 connector. Orange on pin 2 must be mated only with Orange and White on pin 1. No loopback test can determine a miswire. Your symptoms are typically of this wiring mistake.

With Cat6 wire, that connection should still work much faster on that distance. (Actually what is probably happening. Network must resend the same packets repeatedly if the twisted pair are not properly matched - causing bandwidth reductions).

 

[Swampster]

In addition to the 303' from the switch in the control room to the switch in the Sound Booth, it is 38' from that switch to the Welcome Center, which is the farthest computer (341').

In addition to a loop-back test, the connections were tested with a standard cable tester that gives a green LED, or a red LED, or no LED, based on the correctness of the pair. Would this not resolve the question of the pairs being mismatched?

Also, I have seen different punch-down boxes for CAT5 and CAT6, but have not found any different RJ45 terminals for CAT5 and CAT6.

 

[skyking]

yup, that's typically referred to as a "wiremap" test. passing that is a must, but by no means insures that the wiring is up to specifications. A couple of nasty bends or other pinch points can cause a reflection. Scott covered all that in his first post.
CAT5 and CAT6 connectors are identical, providing both ends are terminated the same.

 

[westom]

Originally posted by: Swampster
In addition to a loop-back test, the connections were tested with a standard cable tester that gives a green LED, or a red LED, or no LED, based on the correctness of the pair. Would this not resolve the question of the pairs being mismatched?

Obviously not. Pin 1 could be a brown wire. Pin two an Orange wire. But that tester only knows a wire connects pin 1 to pin 1. Nothing more. Pin 1 must be the orange and white wire. Tester typically would have no idea what twisted pair is and no means to measure it.

You must visually confirm that proper wire assigned as stated previously. If another and easier alternative existed, it would have been posted. Knowing that you used the tester, a visual inspection was posted as necessary.

300 feet is the speced maximum. That means 314 feet will probably work - maybe not always reliable. If the spec says it works at 300 feet, then it must be designed for maybe 360 feet. At those longer lengths, you may have some reliabiltiy problem - minor and trivial. Your's is a complete failure - which is traceable to many problems including mismatched wire pairs. A solution detailed because no other alternative other than visual inspection exists - assuming you do not have $thousands in test equipment.

 

[spidey07]

Originally posted by: skyking
yup, that's typically referred to as a "wiremap" test. passing that is a must, but by no means insures that the wiring is up to specifications. A couple of nasty bends or other pinch points can cause a reflection. Scott covered all that in his first post.
CAT5 and CAT6 connectors are identical, providing both ends are terminated the same.

Not really. cat6 is thicker gauge and the ends keep the twist maintained much closer to the pins. There are significant differences.

Not that anybody should be putting ends on anyway.

 

[skyking]

Originally posted by: spidey07

Originally posted by: skyking
yup, that's typically referred to as a "wiremap" test. passing that is a must, but by no means insures that the wiring is up to specifications. A couple of nasty bends or other pinch points can cause a reflection. Scott covered all that in his first post.
CAT5 and CAT6 connectors are identical, providing both ends are terminated the same.

Not really. cat6 is thicker gauge and the ends keep the twist maintained much closer to the pins. There are significant differences.

Not that anybody should be putting ends on anyway.

RJ45 is..
RJ45

As to putting on connectors, I was not EVEN going there:laugh:

 

[spidey07]

Originally posted by: skyking


RJ45 is..
RJ45

As to putting on connectors, I was not EVEN going there:laugh:

*smacks skyking with a wet trout*

Not when you're talking category specifications it isn't. RJ45 is the form factor/connector.

 

[ScottMac]

Originally posted by: westom

Originally posted by: Swampster
In addition to a loop-back test, the connections were tested with a standard cable tester that gives a green LED, or a red LED, or no LED, based on the correctness of the pair. Would this not resolve the question of the pairs being mismatched?

Obviously not. Pin 1 could be a brown wire. Pin two an Orange wire. But that tester only knows a wire connects pin 1 to pin 1. Nothing more. Pin 1 must be the orange and white wire. Tester typically would have no idea what twisted pair is and no means to measure it.

You must visually confirm that proper wire assigned as stated previously. If another and easier alternative existed, it would have been posted. Knowing that you used the tester, a visual inspection was posted as necessary.

300 feet is the speced maximum. That means 314 feet will probably work - maybe not always reliable. If the spec says it works at 300 feet, then it must be designed for maybe 360 feet. At those longer lengths, you may have some reliabiltiy problem - minor and trivial. Your's is a complete failure - which is traceable to many problems including mismatched wire pairs. A solution detailed because no other alternative other than visual inspection exists - assuming you do not have $thousands in test equipment.

100 meters(328 feet) overall (90M solid core + 5M stranded at each end) is the RECOMMENDATION. The actual spec only describes levels of attenuation, crosstalk, ACR, skew, etc.

By maintaining proper matched components end-to-end, you are "likely" to have a compliant system. The only way to know for sure is to run a compliance test.

300 feet is well within spec and should work every bit as well as ten feet, if properly terminated, and the cabling is in good condition (and also in spec for bend radius, stretch, crush, twist, free of animal pee, etc).

Following the proper pin positions / color code is also critical, at least up to Cat5. At Cat5e and above colors are not functionally critical, but it take no more effort to arrange the colors properly as not, so why not just do it right?