Keith via plug on 8 Feb 2025 09:30:18 -0800 |
[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]
Re: [PLUG] How to diagnose poor Ethernet throughput with Mint |
1) Cat 5 vs. Cat 6... just do Cat 5e (no one should be running cat 5 anymore)
The advantages are just not there anymore. If I want 10Gb or more, DAC cable or fiber is a better option. With DAC you'll even save on the cost of the transceivers because they already come with SFP+ connectors.
2) Shielded vs. Un-shielded... just do un-shielded and pick a better path, even if its longer
I haven't run a shield cable in at least 15 years because such deployments require proper physical grounding to be effective. Both types of cable when used for ethernet are still using differential signaling but as you have found out RFI will still be an issue. Shield cable may help with RFI but its really to protect against electrical induction. For example, I would use shielded cable on T1 lines back in the day so that any charge build up would be kept away from down stream from my equipment. For in-house wiring simply take a path that is away from any electrical as much as possible (but minimum standards). That is easy to do in a data center but it gets more complex in commercial building. The chases are usually full and no one removes old copper. In residential, this can even be more difficult since the vast majority of homes were not built with the concept of a central chase or construction where it is easy to running cable. Not to throw electricians under the bus but those guys often "learned" how to do this stuff on the job. Its not really their thing and that's ok if they are honest about it. If you're not doing it yourself or bringing in a communications contractor, then usually the telco and cable folks you might know can handle simple jobs and have the test equipment. Any decent tester (i.e. fluke) will tell you link speed so always ask to see that. That will be extremely reliable since the tester is actually talking to the far device for link speed negotiation so its more meaning that just a layer 1 (physical) test.
I say all that to say, your situation is not unheard of. However, I would submit that if you think you have to use shielded cable, you're probably taking the wrong path to begin with. If this is just a 1Gb data connection you more than likely will not hit the length limit for reliable communication over cat 5e. You have plenty of line length to work with to find a good path.
Here is something else to consider (and what I would do for a situation where I don't need PoE), if your router has an SFP / SFP+ port or if you have the space to put in a small switch that does, run fiber. Fiber is immune to RFI so it can run along side electrical. We usually put it in a plenum inner-duct (a plastic tube- you've seen them) but for residential you could just use wire modeling of some sort. It might actually be cheaper than running copper and it's not something you'd have to upgrade for very long time if ever.
On 2/8/25 10:26, bergman--- via plug wrote:
In the message dated: Sat, 08 Feb 2025 07:31:22 -0500, The pithy ruminations from Steve Litt via plug on [Re: [PLUG] How to diagnose poor Ethernet throughput with Mint] were: => => On Thu, 6 Feb 2025 21:04:25 -0500 => Frank via plug <plug@lists.phillylinux.org> wrote: => => > Thanks, Walt. => > => > Yes, the problem turns out to be the new cable (cat 6-unshielded) I => > had my electrician run. I ran a very long Ethernet cable through my => > house and got the 1 Gb/s connection I was expecting. My electrician => > hasn't come back to re-run it yet, but after doing some reading, I => > suspect that problem is that we ran it right up against a lot of => > electrical lines in my basement. Neither of us knew it was a => > problem. I'm guessing that after I have him move it, it'll work fine. I submit that any electrician who doesn't know that basic information isn't qualified to run data cabling, and certainly shouldn't be brought back (and paid!) to re-do the work. => > => > I wasn't able to access the National Electric Codes, but random => > Internet web sites say the NEC code is to keep unshielded ethernet => > anywhere between 2" and 8" from parallel electrical lines, but it can => > cross electrical lines at the perpendicular without much of a problem. Yep. Some observations: CAT6 is much less forgiving of terminations, bend radius, and cables being tightly bound (ie., pulling zip-tie cable straps too much) than CAT5* if you're running cable in enclosed spaces, please use plenum-rated cabling (probably overkill in most residential installs, but NEC required & best practice in commercial spaces) electrical interference often manifests as sporadic issues (ie., the problem appears only when the nearby electrical line is under load). Persistent issues with throughput are often the result of poor termination, overly tight bend radius, poor termination, creased cables, poor termination, and, oh, poor cable termination in addition to avoiding close parallel runs to power lines, be cautious about running data cabling near equipment with large, spikey power requirements and large capacitors -- that's not uncommon in mechanical equipment like HVAC systems, water pumps, garage door openers, etc as far as testing the cabling -- any qualified electrician should test the data Layer 1 connectivity before they declare the job to be complete... a really good one will provide a written report of quality/bandwidth before you sign off on the job, rather than simply checking for a pass/fail (ie., connectivity/short, correct/mis-paired) condition => > => > Thanks for the replies (that I wasn't able to reply to). => => Ouch! After reading this, I'm going to make sure to use shielded Cat6, => ESPECIALLY if I'm going to pay an electrician to install it. => => If it turns out you have to pay him again to run it elsewhere, I'd => recommend popping the extra money to have it shielded. Shielded cabling is uncommon and has even stricter requirements for termination -- make sure that your installer is qualified or don't bother. Mark => => SteveT => => Steve Litt => Spring 2023 featured book: Troubleshooting Techniques of the Successful => Technologist http://www.troubleshooters.com/techniques => ___________________________________________________________________________ => Philadelphia Linux Users Group -- http://www.phillylinux.org => Announcements - http://lists.phillylinux.org/mailman/listinfo/plug-announce => General Discussion -- http://lists.phillylinux.org/mailman/listinfo/plug
-- ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ Keith C. Perry, MS E.E. Managing Member, DAO Technologies LLC (O) +1.215.525.4165 x2033 (M) +1.215.432.5167 www.daotechnologies.com ___________________________________________________________________________ Philadelphia Linux Users Group -- http://www.phillylinux.org Announcements - http://lists.phillylinux.org/mailman/listinfo/plug-announce General Discussion -- http://lists.phillylinux.org/mailman/listinfo/plug