Best way to truly test network cables and connections

I have a small client office with about 15 users.  They seem to have data corruption problems from client to server that might be related to networking issues (cables, connectors, switch ports).  Everything test out fine with  a simple pin out tester, but what is a low cost solution to perform a more indepth analysis on each nodes connection?
Who is Participating?
aleghartConnect With a Mentor Commented:
"low cost" is relative.  It's cheaper than other testers, and will fit in your backpack:

Autotests up to 1000BASE-T, test for impedance, crosstalk, wire length, etc.  I use mine even at home when I'm running network & phone cables as well as 75-ohm CATV.  A bit geeky, but better safe than sorry.  Helped me to find an improperly wired patch cable from a FIOS installer.

Great for troubleshooting during construction/retrofit.  Found some cable shorts that could be fixed.  Some that couldn't, but tagged the wires so nobody would try using them for anything but analog phone.

Goes way beyond a blinking light showing continuity.

Lets you blink ports on a unplug computer, plug in tester, blink.  Go to your switches and look for the constantly blinking light.  Helps a lot when jacks are not labeled and/or when you have spaghetti patch cords in the closet.
Check drivers on server? Update to latest?
Duplex setting mis-match on switch?
all clients?
murrycAuthor Commented:
I will check those.

I was thinking maybe there was some software application I could run on my laptop to test a connection or maybe a device that did not cost much that I could purchase to run tests.

So many applications use server side data and corruption is easy when you have bad wiring.
Improve Your Query Performance Tuning

In this FREE six-day email course, you'll learn from Janis Griffin, Database Performance Evangelist. She'll teach 12 steps that you can use to optimize your queries as much as possible and see measurable results in your work. Get started today!

Murryc - A simple cable tester may only verify the first layer (Physical) of the OSI model. As you verify your issue you need to rule out the next layer. For example, your switch may be the next place to start to rule out a layer 2 or 3 issue. For all we now the issue could lie within the OS. What exactly is happening and why do you think it is a network problem?
Murryc if you have a connection and it is passing a test with a cable tester I would feel pretty safe in saying that your issue is not with your wiring.
murrycAuthor Commented:
You address a good point.  There is really no way to know for sure if the problem iss network related, but there are so mnay human variables in the network connection (crimped connectors, poor switched cables, wall plate punches, patch panel punches with wires that might test good on physical layer but one of the conductors is barely hanging on becuase it was sliced with the tool).

It seems it would be really nice to ELIMINATE the network, both physical and other layers so that you can move onto the more dynamic variables in the equation which is the server and workstation.
Fred MarshallConnect With a Mentor PrincipalCommented:
I can't be sure what you mean by a "simple cable tester".  I would certainly use a cable tester that checks that the connection of the *twisted pairs* is proper - not just that there is copper between the proper pins.  The tester should tell you if a couple of the wires are swapped between pairs.  Depending on the length, misapplied pairs can work or not work or be marginal.  They will probably work at 10 feet and not at 50 feet.  Somewhere in between  they could well be flaky.  

You may waste time "fixing" cables that don't meet 568A or 568B.  Contrary to popular belief by some, it doesn't matter if a cable is wired "A" or "B" (or something that's not either "A" or "B" that uses pairs properly).  One can mix and match these cables in a system with no problem (the electrons don't know what color the insulation is).  What *does* matter is rewiring one end of a cable of you don't know what's at the other end.  So, it *is* best to stick to a standard and "B" is it.  That's how all the manufactured cables come wired.
But, if cables don't meet the pairing needs then the tester should tell you and those must be replaced / reterminated.  
Here is a reference:

Also, cables should not be run close to devices that are electromagnetic noise generators nor should they be run in parallel with and in close proximity to power wires.  Best practice would have them run in parallel at some distance.  I've not seen any numbers but I use a minimum of 1 foot in that case.  And, when they need to cross, cross them at right angles.  If unavoidable then longer violations are worse than short ones.

I built a simple cable noise tester with an electronic volt-ohm meter like this:

Meter set to measure AC volts.
RJ-45 connector wired to a short length of CAT5 cable with all 8 wires coming out as short pigtails.
Connect the meter to one of the pairs on the pigtails - alternately to two wires that *aren't* in the same pair.  A more permanent connection would be more convenient to handle.
Connect the RJ-45 connector to a good cable that's out in the open and away from things.
Measure the ac volts as a baseline.
Move the cable around power wiring and flourescent light fixtures to see the ac voltage go up as pickup increases.  Use this experience as another baseline.
Now, plug the test connector into a cable to be tested - with the far end disconnected.
From the ac voltage you see, you'll get an idea if this cable is "quiet" or "noisy" relative to your two baseline numbers.  Maybe you'll get some insight this way.

You seem to say that the problems *only* from/to the server.  Or, might they be between computers as well.  That would be a clue.

It could be worthwhile to purchase a managed switch.  A Linksys SRW208 can be had new for around $140. Replace the server connection to the LAN and a few of the critical computer connections with this switch.  Observe the port errors.  That could be a clue.

If you like, go one step further:
Set up SNMP on the switch and, if available, on the server.  Use PRTG Traffic Grapher (the free version will do for this) on a PC that will be the SNMP "server".   Connect this PC to the LAN and monitor port traffic and port errors.  It works best if you enter into the Port Management what computer/cable is connected to each port on the switch.

You could start with the easiest / least expensive approaches first and go further if necessary.

Fred MarshallPrincipalCommented:
From this:
Replace the server connection to the LAN and a few of the critical computer connections with this switch.
I meant:
Insert the switch *between* the server and other computers and downstream switches so that you have full connectivity (without loops).
To do this you would likely move computer cables from an existing switch to the inserted switch.  As you have more computers than the SRW208 has ports you'd maybe swap computers between the SRW208 and the existing switch(es) to find trouble spots.

Server LAN ...  SRW208 .... some computers and existing switch .... the rest of the computers.
murrycAuthor Commented:
The cable noise tester is a perfect example of what I am referring to.  A physical cable tester will test existence of wires and proper paring as you mention, but being able to test the noise issue as well as actual data transfer back and forth, this is what I really need.  Like I mentioned, I have run into conductors that were sliced and barely hanging on.  They passed the cable tester, but when they were in use, they could not actually pass data cleanly - dropped packets.

So there is a need to test switches, servers and workstations, but the need to fully test the cable, connectors, wall plates and patch panels is what I am wanting to solve.

Related to your thoughts on the placements of cables withint distance of electrical lines.  That is all correct, but actually knowing whether the cable runs were successful in avoiding all of that is near impossible without some level of testing after the fact.

Additional thoughts on the perfect low cost tester for these needs?
itnetworknConnect With a Mentor Commented:
Murrayc - I would start with getting the Fluke that aleghart mentioned. It will be a invaluable tool to test the physical layer. If you are having issues troubleshooting these type of issues I would also recommend that you purchase a managed switch as fmarshall has. I would recommend steering away from a Linksys switch. If cost is an issue to you, a 24 port 2950 series can be had as a refurb w/ a warranty for anyware around $180 to $225 from a reputable seller. With a Cisco switch you will have more versatilaty in troubleshooting and will be using a solid "commercial" solution for your company.

My main point that I would like to restate is that many times in my working experience when I have heard of similar issues, and that the suspicion was on cabling that was passing traffic, that the problem really ended up being another unrelated OSI layer problem i.e.; server, client, software, routing, switch config issue, bandwidth issues, etc. I agree that you need a tool like the Fluke model stated to properly feel confidant that you can work your way past troubleshooting the physical layer of this problem.
For a free software tester, consider WireShark:
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

All Courses

From novice to tech pro — start learning today.