New house - ethernet/networking advice please!

adamCCC

Established Member
Joined
Aug 27, 2009
Messages
311
Reaction score
13
Points
98
Age
35
Hi

Hoping to get some networking advice for my new house, we'll most likely live here for a long time so future proofing as much as possible is important.

Currently have BT FTTP internet which terminates in the living room. The plan is to run ethernet from the fibre connection to a networking cabinet (with a switch - TP-Link 8-Port Gigabit Ethernet Switch, hue bridge etc) upstairs on the landing (central to whole house - house is a weird layout).

Given we'll live here for the long term im thinking cat6a is a good choice. From googling it seems that shielded cable is unnecessary however I can only really find shielded cat6a cables.

In my previous house (6 years ago) I got cat7 cable (it was a lightning deal on amazon and really cheap) and terminated it into faceplates - please see pics. Since I've been googling alot around this I'm now starting to think I didnt even do it right back then! Was it the wrong type of face plate? Was using the copper tape to hold back the braid correct?

In particular, im confused about grounding in regards to shielded cable - there seems to be alot of differing opinions about this!

2 options for new cable are:

External SHIELDED CAT6A Outdoor Use COPPER Ethernet Cable S/FTP Reel 50m Black

Oren CAT7 50M Ethernet Cable - 23 AWG Pure Copper Wire - B2ca - 900 MHz Bandwidth S/FTP Internet LAN Network Cable, Halogen Free, Non-Corrosive, Flame Retardant – 10GbE High Speed Ethernet Cable (wont let me post the amazon link)

Could someone please clarify the right way to install this new cable. The first job is to install a faceplate in an upstairs office room.

Any help will be much appreciated!

Thanks
 

Attachments

  • 20180328_171626.jpg
    20180328_171626.jpg
    79.1 KB · Views: 104
  • 20180328_174727.jpg
    20180328_174727.jpg
    88.7 KB · Views: 106
  • 20180403_152609_HDR.jpg
    20180403_152609_HDR.jpg
    179.1 KB · Views: 105
  • 20180403_184735_HDR.jpg
    20180403_184735_HDR.jpg
    133.6 KB · Views: 127
  • 20180403_193153_HDR.jpg
    20180403_193153_HDR.jpg
    103.9 KB · Views: 109
Last edited:
The copper tape is part of the shielding; it's effectively a double shield.

As CAT6a usually has some shielding, this should be appropriately terminated too, as otherwise, you might run into issues. You might never get or notice any issues as you had in your previous house but best to do it right in the new one if you are starting from scratch.

If you look at CAT6a modules then you should notice they have a metal ground connection as well as shown below.

1649842207508.png


Equally CAT6a patch cables will have a metal shield on the plug itself.

Depending on the switch, it will also potentially have a metal RJ45 socket so that there is a continuous shield to the device. The device might well have a plastic RJ45 socket so the sheild doesn't actually have an electrical shield connection at that point. That's like some hifi devices, where the shield is only connected to ground at one end. What you don't want is bits of the run with a shield not connected to ground.
 
Given we'll live here for the long term im thinking cat6a is a good choice. From googling it seems that shielded cable is unnecessary however I can only really find shielded cat6a cables.

With the caveat that I haven't read any of the TIA "cat" standards in years; IIRC the standards themselves do not mandate the construction of the cable or the use of shields (though they do mandate the use of pure copper conductors - no CCA) but they instead describe the electrical performance in terms of things like frequency response, cross talks, etc. and leave it to the manufacturers as to how they achieve that.

The higher cats have higher specs, (and more of them) and to achieve that a lot of manufacturers opt for things like shields, mechanical dividers and so on. The up shot being that most higher cat cable has shields and dividers to achieve the required electrical performance. Some even has shields around the individual pairs and an overall shield (and a drain wire.) As Chuck says, all of that needs to be earthed. In commercial installations where we are using dedicated comms. cabinets/rooms, we usually tie everything to earth in said comms. cabinets/rooms. Appropriate patch panels and racks often have earthing posts (and internal pathways) to achieve this.

Could someone please clarify the right way to install this new cable. The first job is to install a faceplate in an upstairs office room.

At some cat or other - it may be 6 or 6a - the standards mandate than cables must be installed into "proper" "containment" - trays, conduits and so on - so poking things behind walls and clipping it to a handy pipe (or whatever) instantly "fails" it as cat whatever on visual inspection; in much the same way that a rusty car can fail it's MOT on a visual inspection even though all the mechanical tests are a pass and it's otherwise fuctioning adequately as a "car."

But that doesn't mean your cables won't "work," even without putting on a very expensive certification tester (they cost hundred/thousands) to run the battery of tests required to certify each lobe at a given "cat." Gigabit ethernet is well within the performance limits of even cat5e and you have to do a spectacularly bad job of installing the cable (bad termination is the worse culprit) for gig ethernet to not "work." Just don't be disappointed in years to come if you try 10Gig ethernet and it doesn't work (or more likely if the transceiver market goes as I hope, falls back to 5Gig or 2.5Gig.)

I am fond of saying, there's a lot more to achieving the higher "cats" than what cables and connectors you buy - though that's of course the starting point. For the higher cats, the installation standards (methods) are increasingly stringent.

...

Is there anything specific you want to know in this questions...? Generally, one establishes the cable routes, installs the containment (if you are using any,) lay in the the cables, terminates them and tests them (in pretty much that order.) For a DIY install you might be omitting some of that (such as the containment and testing.)

Don't forget to label it all. It's surprising how fast one forget what goes where and of course it's really useful for a future owner. Just ensure each lobe is labelled both ends with a unique identifier. If you wanted to go really "professional" you could record it all in a document (some even have diagrams) describing what goes where (reflecting the unique ID of each lobe) what cable was used, when the work was done and how it was tested/certified - even if it's just a basic continuity test - it's also useful information for future owners.

Do your best to not kink, crush or nick the cabling as you lay it in. So doing can compromise the electrical performance and eat in to the signal margins a bit. Any changes of direction should be "curved" and not "bent" - so no "hammering" around 90 degree corners. IIRC minimum bend radius is about 4x the sheath diameter.

AVF Mantra is to "always install two" (or more) along any given route. It's highly unlikely UTP will fail in service, but if one does, depending on the cause of the failure, if you have an alternate in situ, you stand a chance of getting something up and running again fairly quickly. Without an alternate, you are off the air until you rip and replace. Cable is cheap compared to the time/hassle of installing it.

Here's my favourite DIY site for UTP - it has lots of useful diagrams and such...

 
Last edited:
I think one of the problems with cable categories is people don't understand either the cables and what they do or why they are done that particular manner. On the whole, the biggest problem the various cat's try to combat is interference of one type or other, and if none exists in the ambient environment or are not prone to the effects then you're largely wasting your time with more complex and expensive cabling.

Shielding done incorrectly tends to cause way more problems that no shielding at all, and in fact the higher tolerances and transmission rates are gained primarily by increasing the number of twists per cm and not by the shielding. In domestic occupations I cannot see the point of shielding at all, and unless there is a specific need it should be entirely avoided. The belief there there is a need to continually scale the versions of CAT for network speed is mind boggling. Even CAT5e will run 10gb up to 45Meters which I'm pretty sure is within scope of most domestic households. Future proofing? possibly, but I suspect we'll be long done with copper by the time anything past 10gb will be domestically available.
 
I have opined previously in these columns that; given the massive installed base of cat5 and cat5e in all the worlds office blocks, which will be hugely expensive to replace, there is surely a commercial incentive for the clever people to invent new ways to use that installed base to deliver higher speeds as/when/if it becomes necessary. Viz: if new NIC's and new switches costs X (and takes a few relatively low disruptive outages to install) and a recabling exercise cost multiples of X and takes a few weeks/months of people pulling bits of the building to pieces, I guess it is no surprise which is going to be most attractive to the business bean counters.

Equally, I can see that if the cost difference of a DIY install of catX versus catY is minimal one "might as well" go for the highest, especially if one is not going to observe all the installation stipulations and not test/certify it and thusly, strictly speaking, the completed job is not "cat" anything. But I would avoid anything shielded as earthing all the shields adds complexity (now) for no gain (now.) UTP is more than good enough Gigabit ethernet, and at the shorter distances of SOHO cable lobes, as Jim says, is probably going to be fine for 2.5/5/10Gig unless one does a seriously poor job of the installation.

Anyone truly "worried" about "future proofing" their SOHO infrastructure would be best advised to install whatever they selected into "proper" containment (ducts, conduits, trays, trunking, etc.) so that if/when the need comes for somethings faster, if the incumbent solution could not hack it (or does not work) it's an easier job to replace it than if it's chased into channels and plastered over (which incidentally fails Cat in any case) clipped to noggins, tucked under skirting boards, poked behind plaster-boards and so on.
 
Last edited:
Based on current and projected use cases 10Gb will be plenty, even a 1Gb is plenty for the vast majority of people and 100mb to the desktop is pretty standard in the office. Sure we have some engineers running at 10Gb but you could probably count those in the hundreds out of 100,000+ users. Remember the device they are connecting to needs to serve multiple users as well.

You have to remember that app developers know most users don't have more than 100mb so why design something that needs that much data bandwidth, so keep processing to source and only transfer processed data. And more and more people using using WiFi.
 
1GB is plenty for most users... esp as most SOHO devices cant feed data much faster than that (or are the bottleneck in themselves.)

You need to ask yourself, what is the long term potential for exceeding 1Gb? Whilst 10Gb is coming down in price, can be found on increasing number of motherboards, unless you have a shared storage device with 10Gb interface then you are probably not going to need 10Gb in the medium term.
And when you do, you will need a switch that supports it. Which gets very expensive, very quickly, as they generally require SFPs

"100mb to the desktop is pretty standard in the office"? I think the 2000's want their standards back then, as that was pretty much standard in the late 90's, I haven't seen an office using FastEthernet in well over a decade and a half.
I'm not sure the last time I saw a device that topped out at 100mb.
 
"100mb to the desktop is pretty standard in the office"? I think the 2000's want their standards back then, as that was pretty much standard in the late 90's, I haven't seen an office using FastEthernet in well over a decade and a half.
I'm not sure the last time I saw a device that topped out at 100mb.
Pop over to any large office with thousands of users, very few are using Gb. Partially cabling but mainly switch costs.
 
Pop over to any large office with thousands of users, very few are using Gb. Partially cabling but mainly switch costs.

It depends on when they were last rewired. A number will have been rewired for VOIP and have had CAT6 rewire. Depending on the company involved then the switchgear will often be gigabit, whether it has been enabled as such depends on the needs. Also seen it where the devices have been daisy-chained off the VOIP devices and they only have 100Mbit pass through.
 
For the avoidance of doubt, VOIP does not require any particular "cat" cable. It may just be coincidence in the example Chuck cites that "extra" UTP needed to be installed to avail the VOIP handsets and they just happened to select CAT6 while they were about it.

I did a big VOIP deployment a few years ago and it worked fine over cat5/5e (and largely 100mbps at that time.) The bandwidth for VOIP is tiny - I could bearly measure it on the Mitel system I deployed.
 
Last edited:
For the avoidance of doubt, VOIP does not require any particular "cat" cable. It may just be coincidence in the example Chuck cites that "extra" UTP needed to be installed to avail the VOIP handsets and they just happened to select CAT6 while they were about it.

I did a big VOIP deployment a few years ago and it worked find over cat5/5e (and largely 100mbps at that time.) The bandwidth for VOIP is tiny - I could bearly measure it on the Mitel system I deployed.

Yep those office infrastructures had "ropey" existing CAT5e cables and rather than try and potentially bodge the cables and install, companies took the decision to rerun all cables as part of the VOIP rollout. As you say VOIP doesn't need a particular CATx cable, it just happened to be CAT6 chosen in this case for longetivity.
 
It depends on when they were last rewired. A number will have been rewired for VOIP and have had CAT6 rewire. Depending on the company involved then the switchgear will often be gigabit, whether it has been enabled as such depends on the needs. Also seen it where the devices have been daisy-chained off the VOIP devices and they only have 100Mbit pass through.
We did VoIP a few years ago in our building, still got the old cabling as it was too expensive to pull out. I've worked in a couple of building with couple of thousand people, some rewired and as I mentioned while many are cabled to support 1Gb and as we slowly upgrade switches we are changing to 1Gb but there isn't any rush. Switch side its more that you can't get 100mb line cards for new switches which is the driving factor.

Which ties into consider server side virtually everything is virtualised and even with 40Gb or 100Gb on ESX hosts with 50+ VM on there they can't keep up hundreds of connections per VM giving thousands of connections per host. We still do petabytes of NAS but more and more often the data is stored in some sort of database or some other structured storage so access is via servers on VM.

There are use cases were we do 1Gb or even 10Gb to the desk but anybody dealing with that volume of data for example with simulation, CAD, ML, Matlab or whatever 99% of the time the processing is taking place server side. This is especially even more true these days when so many people are WFM and VPN in so if high speed network is required for client its easier to setup a VDI session where the client is in the data centre as the data.

For SOHO or SMB, Gb networking is not much more expensive and easy but in the enterprise, there is limited advantages vs cost.
 
Where the switches PoE, that was the major upgrade for and went gigabit at the same time.
 
Where the switches PoE, that was the major upgrade for and went gigabit at the same time.
When we rolled out PoE we were still using Catalyst 45xx where there was a price premium for 1Gb line cards. Now using Catalyst 9xxx range, I don't think there are mb line cards even available.
 
The mainframe is dead - long live the mainframe..! :D
Lol, I was in a presentation meeting last month were our mainframe team was excited that they are upgrading to Cobol v6.

Anyway, review from Linus, I'm pretty sure the $1m is a starting point.

 
Anyway, review from Linus, I'm pretty sure the $1m is a starting point.

Nah, that's what you have to pay for a quote ;)

When we rolled out PoE we were still using Catalyst 45xx where there was a price premium for 1Gb line cards. Now using Catalyst 9xxx range, I don't think there are mb line cards even available.

Yep we didn't go with Cisco as we were a competitor :D
 
Pop over to any large office with thousands of users, very few are using Gb. Partially cabling but mainly switch costs.
That would be every company I have worked with or consulted to in the last 15 or so years, or any of the customers of my current employer. Even the last SME I worked with 10 years ago migrated to Gbe. The last large scale company on FE was 2004, and they were doing a 12000 seat migrating to Gbe to the desktop.
 
When we rolled out PoE we were still using Catalyst 45xx where there was a price premium for 1Gb line cards. Now using Catalyst 9xxx range, I don't think there are mb line cards even available.
Not for 9200/9300, the ports are 1gb or 10gb :)
 
Not for 9200/9300, the ports are 1gb or 10gb :)
Yeah, that is the driver for our swapover. Very few people upgrading to 1Gb specifically, just happening as a side effect of tech refresh. And tech refresh could be 5 to 7 years. Just had a quick look at some old orders and back in 2016 there was a 30% cost increase for 1Gb PoE line card Vs FE for a catalyst 45xx switch.

Tbh, more users are connecting via WiFi these days.
 
Last edited:

The latest video from AVForums

Is 4K Blu-ray Worth It?
Subscribe to our YouTube channel
Back
Top Bottom