Fiber vs Copper

[aefriot]

We are in the process of purchasing a home that is in need of a complete electrical rewire. In the process, we want to make sure the home is cabled for our network and security. The walls are adobe and we are not sure the home will be good for wireless because of the walls and the layout (Included). We wish to do this only once. We work from home and have a multitude of hobbies and need access to the network for many of them. We are in our 50's and do not wish to do this again. Due to the layout of the home, I can see no way around having an access point in every room as I cannot rely on good wireless service through the adobe walls. Then, the distance between rooms may be unrealistic to use wireless between rooms even with walls easily penetrated by WIFI.
We will likely have a solar installation in the rear of the property and will need WIFI for status reporting behind the back building.

Knowing we don't really understand the requirements of the future, in 20 years, might the need for bandwidth go beyond copper?

Considering the number of access points, would costs be prohibitive to install fiber throughout the home. I would need to purchase as many as fifteen access points. Each with it's own converter. Are there any single board computers w/ WIFI/Bluetooth, Zigbee and fiber? Is there an open mesh networking software solution for home users that could be incorporated in the network?

Maybe someone else has experience with similar circumstances in home or professional experience?

I wouldn't mind experimenting with developing a system for use with smart home devices as now would be a great time to integrate them while we are in the re-construction phase and running wire anyway.

Thanks.

 

[Rock]

15 access points?

Consider that the roof is probably more easily penetrated by the Wi-Fi signal than the adobe walls. If I am right, then putting the AP's high outside can cut your count dramatically.

 

[etorix]

Knowing we don't really understand the requirements of the future, in 20 years, might the need for bandwidth go beyond copper?

Prediction is notoriously difficult, especially about the future.
Copper stops, for good, at 10 Gb/s, using quite some power for extended lengths. You may not need faster than that.
But fibre uses less power and cheaper switches (compared with copper 10G), which are valuable benefits even without considering that it would be ready for 25 Gb/s, 50 Gb/s, etc. by changing the transponders.

This looks like a prime opportunity to ask for a quote for running fibre through these walls, either alone or alongside some copper for compatibility with 1 Gb/s devices and/or PoE(++).

 

[CyklonDX]

in ref. to original question

Real cons/pros of fiber vs copper

copper if you want to do sharp edges
copper is still passing through electricity so keep that in mind - interference from power lines can be felt (especially if you used not up-to-std cables, and they run close and are 30-40A cables.)
copper cable quality, most of them are under their spec quality; You can often meet unpleasant reality that the cable you bought is actually cat5e internally and not cat6-8.
copper cabling is durable, and even if you have a problem with rodents or animals like cats/dogs - it will take them significant amount of biting before its done.

fiber you can't bend it too much as you will either loose signal or break the cable all together *there are types of fiber cables that can bend more but typically also offer worse performance.
you can have fiber lines together with your power lines.
its just light, so if you have a slight damage to coating it may still work - and won't cause fire or power leaking etc.
fiber is typically universally good. If it can reach the other side you will 99% of time won't have any issues of it not performing up-to-spec.
a small bite from rodent/cat would ruin the cable unless you bought fiber within proper protection.

In terms of changing transponders etc, that will only work if your nic is also rated for those speeds. Sure it offers more variety; but on copper you also can push 25-40Gig too over cat8 cables *(fiber will be more expensive in the end).

 

[Scott Laird]

Guidelines:

• If you need to connect between different electrical systems with different grounds, then always use fiber. Between buildings should ~always be fiber.
• If you need > 10 Gbps, then you'll need fiber. *Technically* there are specs for 25G and 40G over copper, but there is (and has never been) any commercial hardware for either, and they won't work with Cat6-ish cable anyway. So just ignore it and assume that we'll never see >10G on copper for reasonable prices.
• If you want to run power over the same cable as data, then you need copper.
• If bandwidth isn't an issue, then copper is ubiquitous on devices.

So, for connections to remote buildings, use fiber.

For indoor WiFi APs today, use copper, because PoE makes life easier. 5-10 years down the road, you *might* want copper (for power) and fiber (for bandwidth).

The best move is to install conduit everywhere, then you can retrofit whatever makes sense in the future. Failing that, you'll need a crystal ball. Pull fiber (singlemode) where you think you'll want >10G eventually. Ideally you'd pull 1-2 12-fiber MPO/MTP trunks between each location, but that adds up annoyingly quickly. Failing that, pull at least 2 duplex LC singlemode cables. You can get pre-terminated fiber cut to length, and it's not particularly more expensive than copper, but it's harder to fix if it's damaged.

 

[Scott Laird]

Also, while fiber is less durable / less flexible than copper, it's not really *that* much worse. The bend radius for most fiber is usually just a cm or two, and you can buy armored fiber that is probably more durable than "normal" cat 6. I killed a MPO->4xLC cable last year when I pinched it with a cable management tray's lid, so it's absolutely possible, but it's not *that* easy. The biggest problem is that it's easy to get (especially single-mode) fiber dirty, and you need to clean it before plugging it in, or you'll get signal loss.

 

[kapone]

We are in the process of purchasing a home that is in need of a complete electrical rewire. In the process, we want to make sure the home is cabled for our network and security. The walls are adobe and we are not sure the home will be good for wireless because of the walls and the layout (Included). We wish to do this only once. We work from home and have a multitude of hobbies and need access to the network for many of them. We are in our 50's and do not wish to do this again. Due to the layout of the home, I can see no way around having an access point in every room as I cannot rely on good wireless service through the adobe walls. Then, the distance between rooms may be unrealistic to use wireless between rooms even with walls easily penetrated by WIFI.
We will likely have a solar installation in the rear of the property and will need WIFI for status reporting behind the back building.

Knowing we don't really understand the requirements of the future, in 20 years, might the need for bandwidth go beyond copper?

Considering the number of access points, would costs be prohibitive to install fiber throughout the home. I would need to purchase as many as fifteen access points. Each with it's own converter. Are there any single board computers w/ WIFI/Bluetooth, Zigbee and fiber? Is there an open mesh networking software solution for home users that could be incorporated in the network?

Maybe someone else has experience with similar circumstances in home or professional experience?

I wouldn't mind experimenting with developing a system for use with smart home devices as now would be a great time to integrate them while we are in the re-construction phase and running wire anyway.

Thanks.

On a tangential note...interesting layout of the property. I'd hazard a guess that this isn't in the United States?

Anyway, interesting problem. The adobe walls do pose a challenge with good wifi. I second the thoughts from others above, do it once, do it right and don't do it again. Running wires properly is more expensive than the cost of the wire itself. Some thoughts:

1. Decide on a location for your network/server closet. You may not think you'll have "servers" there, but you never know.

• Run good power to this closet. I'd hazard that running two 20a circuits (110v or 220v, depending on where you're located, but even in the US running 220 is easy) should last the lifetime of the property. If at all possible, I'd choose 220v from the get go and buy equipment based on that.
• Plan for some cooling for this closet. You're gonna need it.

2. Choose a core switch. This is going to help you choose the right downstream equipment, including wiring.
3. Run 2x copper and at least one fiber run to each room. Just do it once, even if you don't think you'll need it. Bring the wires all the way to the network closet. This will expose issues if some runs are too long etc.
4. For your solar installation in the back, there are many "outdoor" access points that are weather resistant and can be mounted outside (with wiring coming to the same network closet). This AP does not need serious horsepower if it's only supporting the solar installation.
5. Access points for the rooms.

• Choose APs that can be run over PoE, it just makes life easier (like @Scott Laird said).
• Choose APs that can "mesh", so that your clients have a single SSID as far as they are concerned.
• A gigabit-ish AP in every room is way more than enough. These are not even that expensive nowadays. Ruckus APs are excellent and they mesh and can be had for ~$30 a pop (Pardon Our Interruption...)

6. Put it all together (is gonna be fun) and enjoy!

 

[reasonsandreasons]

Building off what @Scott Laird said (and apologies if this is taking us off topic) I've always considered fiber a "it works or it doesn't" thing, particularly with bend damage. Is that actually the case, or are there gradations?

 

[fohdeesha]

Building off what @Scott Laird said (and apologies if this is taking us off topic) I've always considered fiber a "it works or it doesn't" thing, particularly with bend damage. Is that actually the case, or are there gradations?

it's 1's and 0's, they either come through correctly or they don't. If the connection is marginal, then it might not present as a completely dead link, but you'll see a lot of framing errors/ CRC errors on the link layer (which are very obvious and easy to spot). In higher bandwidth protocols, 100g+ etc, you have extra things like FEC that come into play

armored singlemode fiber that will run 100gE without a sweat in house distances is like, pennies these days. no reason not to run it to be future proof for the next 50 years if you have the opportunity

ps 15 APs for barely 7k sqft is *insane*

 

[Scott Laird]

Building off what @Scott Laird said (and apologies if this is taking us off topic) I've always considered fiber a "it works or it doesn't" thing, particularly with bend damage. Is that actually the case, or are there gradations?

Most managed switches will report the send/receive power levels for each SFP/QSFP lane; presumably it's possible to bend fiber and hurt its ability to carry light without dropping the received light level so low that it stops working, but I've never really tried. *Generally*, low light levels (for short fiber, at least, I don't manage any multi-km fiber runs) mean that there's dust in one of the connections. Get a fiber cleaner (I use this one from fs.com), clean the connections, and *usually* everything will be back to normal. If not, swap patch cables and/or optics.

 

[mattventura]

Agree that just running copper + fiber to each room is your best bet.

Copper has its advantages, like PoE, but fiber just runs circles around it in every other regard. At 10GbE+, it's cheaper thanks to secondhand transceivers, and uses less power.

Your best fiber option is probably LC duplex SMF. Since there is so much of it buried, there will continue to be a market for faster transceivers for the foreseeable future.

Building off what @Scott Laird said (and apologies if this is taking us off topic) I've always considered fiber a "it works or it doesn't" thing, particularly with bend damage. Is that actually the case, or are there gradations?

I have had one specific model of patch cable degrade to the point where it was starting to lose signal, but it was probably just low quality. They weren't getting bent or impacted, it was running from a switch on a desk to a wall jack behind the desk.

 

[klui]

There are more benefits to copper than PoE.

• Easy termination: If things break it's much easier to reterminate/patch.
• Compatible: Copper is much more available than fiber for many devices and systems.
• Strength: Although fiber has good tensile strength through its protective aramid yarn, it has poor compression/impact strength. Many don't think twice dropping a coil of category cable on a hard floor. Do that with fiber, especially multi trunk assemblies, and one risk of damaging the cable. Even a single MPO-12 while rated for 100 N tensile load, rates ~0.5 N-m and 4 N/mm for impact and compression loads. Because of this weakness it's best practice to deploy fiber through conduit.
• Ease of use: Best practice for using fiber is inspect, clean if necessary then repeat. It applies to new cables as well as used. Can't forget the transceivers/cassette jacks. For hobbyists it's probably fine to blind-clean the tips as good inspection scopes are costly especially for MPO tips. In a highly dusty environment like an unfinished attic or garage, dust can creep into the interconnects over time. One can get carried away getting all the tools and paraphernalia for fiber cleanliness. Much simpler with copper.

 

[aefriot]

15 access points?

Consider that the roof is probably more easily penetrated by the Wi-Fi signal than the adobe walls. If I am right, then putting the AP's high outside can cut your count dramatically.

Likely, yes. We are also planning a radiant barrier to reduce heat penetration/loss in the future. That will likely reduce the WIFI penetration. I did not mention that in the post. I am sorry, but I had not thought of roof penetration for WIFI. That reminds me that the floors are not a barrier to WIFI, therefore cutting the number of access points. That, I hadn't thought of...silly me. Thank you.

 

[aefriot]

Prediction is notoriously difficult, especially about the future.
Copper stops, for good, at 10 Gb/s, using quite some power for extended lengths. You may not need faster than that.
But fibre uses less power and cheaper switches (compared with copper 10G), which are valuable benefits even without considering that it would be ready for 25 Gb/s, 50 Gb/s, etc. by changing the transponders.

This looks like a prime opportunity to ask for a quote for running fibre through these walls, either alone or alongside some copper for compatibility with 1 Gb/s devices and/or PoE(++).

While shielded copper should be immune to electrical interference when run beside power, fiber is definitely immune. When electric is run, it is definitely time to run a digital medium, we are trying to determine our budget. Also, I am accustom to working with copper. Fiber I have spliced and terminated years ago when I was installing fiber cross-country. But not in the last 20 years have I touched it. We have never dealt with higher speeds than gigabit for the home, but the future is certainly going to require higher speeds. How much higher, we don't know. How much we want t increase speeds, we don't know what will be required, yet. 20 years is a long time into the future to look when technologies change fast. Perhaps we should just pick a speed and make it a priority. I don't suppose 10G will be too much to shoot for, as you say, upgrading is a change of the transponder. The upgrade will be cheaper in the future. 10G fiber is likely affordable so we do not have to upgrade in the next 10 years. Will we go beyond the need for 40G? I hope not, though I think we are going to run conduit for the power and fiber side by side and can pull an additional string to pull a new cable if necessary. Cheap enough to do that now to save in the future. We have enough to do in the house and might as well plan for the future at reasonable speeds. Then hope we don't need to upgrade sooner.

 

[aefriot]

in ref. to original question

Real cons/pros of fiber vs copper

copper if you want to do sharp edges
copper is still passing through electricity so keep that in mind - interference from power lines can be felt (especially if you used not up-to-std cables, and they run close and are 30-40A cables.)
copper cable quality, most of them are under their spec quality; You can often meet unpleasant reality that the cable you bought is actually cat5e internally and not cat6-8.
copper cabling is durable, and even if you have a problem with rodents or animals like cats/dogs - it will take them significant amount of biting before its done.

fiber you can't bend it too much as you will either loose signal or break the cable all together *there are types of fiber cables that can bend more but typically also offer worse performance.
you can have fiber lines together with your power lines.
its just light, so if you have a slight damage to coating it may still work - and won't cause fire or power leaking etc.
fiber is typically universally good. If it can reach the other side you will 99% of time won't have any issues of it not performing up-to-spec.
a small bite from rodent/cat would ruin the cable unless you bought fiber within proper protection.

In terms of changing transponders etc, that will only work if your nic is also rated for those speeds. Sure it offers more variety; but on copper you also can push 25-40Gig too over cat8 cables *(fiber will be more expensive in the end).

I use to install fiber optic cross country. I have also spliced fiber, though that was with an actual electric arc and not the more convenient mechanical methods. Yes, bends can be a problem, but I plan on running conduit throughout the house. Install conduit once and pull many different cables many times without tearing out any walls, floors or ceilings. I certainly know what it takes for a rewire and do not wish to make it difficult for myself or others in the future. I also leave extra string to pull new cables later. Bends will always be supported and hopefully, rodents may not be a problem. If they are, pull another cable if the string is good. If not, blow another string through or push the string with a tape. Nothing is 100%, we just work toward making things easier for ourselves in the future.

I thought I read cat8 as not really capable of 40G speeds. In lab testing only, maybe? Interference would still have to be kept to a minimum. Interference is not an issue with fiber as long as there are no sharp bends. Even with copper, I would have to upgrade the equipment, the same as fiber, going from 10G to 40G. Is one going to get cheaper over the other in the future?

 

[aefriot]

Guidelines:

• If you need to connect between different electrical systems with different grounds, then always use fiber. Between buildings should ~always be fiber.
• If you need > 10 Gbps, then you'll need fiber. *Technically* there are specs for 25G and 40G over copper, but there is (and has never been) any commercial hardware for either, and they won't work with Cat6-ish cable anyway. So just ignore it and assume that we'll never see >10G on copper for reasonable prices.
• If you want to run power over the same cable as data, then you need copper.
• If bandwidth isn't an issue, then copper is ubiquitous on devices.

So, for connections to remote buildings, use fiber.

For indoor WiFi APs today, use copper, because PoE makes life easier. 5-10 years down the road, you *might* want copper (for power) and fiber (for bandwidth).

The best move is to install conduit everywhere, then you can retrofit whatever makes sense in the future. Failing that, you'll need a crystal ball. Pull fiber (singlemode) where you think you'll want >10G eventually. Ideally you'd pull 1-2 12-fiber MPO/MTP trunks between each location, but that adds up annoyingly quickly. Failing that, pull at least 2 duplex LC singlemode cables. You can get pre-terminated fiber cut to length, and it's not particularly more expensive than copper, but it's harder to fix if it's damaged.

The ceilings of the structure are 10 feet, I planned to lower that by a foot, install radiant barrier, radiant heat/cooling tubing and electrical and networking all in the ceilings and still have 9 foot ceilings. I am hoping that the foot of space will be enough to allow for lots of room for renovations and interference-free upgrades. Being thick adobe, I am hoping rodents will not be a problem assuming the new conduit will be sealed properly the whole way through the walls. I am familiar with conduit as I placed underground fiber cross country and in buildings within cities. I have not worked with fiber for home networking. All power being in the ceilings will also make ceiling APs easy as power is right there anyway. No POE required, really as cable will be run through the same conduit and be accessible. Each room will have a junction box for both power and network in the ceiling behind a lighting fixture...providing code will allow. I'm not familiar with code here in Spain and am constantly learning differences between the States and Europe. It is interesting.

 

[aefriot]

On a tangential note...interesting layout of the property. I'd hazard a guess that this isn't in the United States?

Anyway, interesting problem. The adobe walls do pose a challenge with good wifi. I second the thoughts from others above, do it once, do it right and don't do it again. Running wires properly is more expensive than the cost of the wire itself. Some thoughts:

1. Decide on a location for your network/server closet. You may not think you'll have "servers" there, but you never know.

• Run good power to this closet. I'd hazard that running two 20a circuits (110v or 220v, depending on where you're located, but even in the US running 220 is easy) should last the lifetime of the property. If at all possible, I'd choose 220v from the get go and buy equipment based on that.
• Plan for some cooling for this closet. You're gonna need it.

2. Choose a core switch. This is going to help you choose the right downstream equipment, including wiring.
3. Run 2x copper and at least one fiber run to each room. Just do it once, even if you don't think you'll need it. Bring the wires all the way to the network closet. This will expose issues if some runs are too long etc.
4. For your solar installation in the back, there are many "outdoor" access points that are weather resistant and can be mounted outside (with wiring coming to the same network closet). This AP does not need serious horsepower if it's only supporting the solar installation.
5. Access points for the rooms.

• Choose APs that can be run over PoE, it just makes life easier (like @Scott Laird said).
• Choose APs that can "mesh", so that your clients have a single SSID as far as they are concerned.
• A gigabit-ish AP in every room is way more than enough. These are not even that expensive nowadays. Ruckus APs are excellent and they mesh and can be had for ~$30 a pop (Pardon Our Interruption...)

6. Put it all together (is gonna be fun) and enjoy!

Yes, I am in Spain. The home is interesting and will be a challenge, but we are up for the tasks. I wish to have a server closet/room. Likely I will have to place it where the solar equipment will be. Power will always be there an on. We are 230V always, unless I also wish to have 3ph in the workshops. But still, 230 will be available in the equipment room for the solar and servers. We wish to have file and terminal server for the dummies in each room. Except for the office where my wife will be working. She requires MS where the rest for the home will be remote DebianOS (likely).
I had not thought of running each a fiber and copper too. It is an interesting thought. If the server room is in with the solar, the longest single cable will be roughly 30 meters. Right at the limit of Cat8. Again, as I said earlier post, is the 40G speeds real? Not just lab speeds? I still haven't seen anything where I am positive which way I am going to go. Fiber or copper (Cat8). If I run both, I can start on copper then upgrade speeds or switch to fiber as equipment becomes less expensive. I guess I haven't seen anything that definitively says on or the other. I appreciate the advice, really. I need to look at hard costs, look at their limits and how long they may last before replacing is necessary. Shoot, there is nothing to say if I buy now for the next 10 years, the equipment will even last that long and I will have to replace it before then anyway. Maybe you are right. Pull both while the work is being done. Buy good equipment with hopes to get 10 years out if it, then upgrade as necessary when equipment prices come down. No need to pay through the nose for high quality extra fast speeds when life expectancy is 10 years only and not yet even accessing its potential. Meaning, not go gung-ho and wasting money that would never be utilized. I think we likely can work with cat6 speeds for 10 years then switch to fiber as needed. Moving equipment around to maximize what equipment is remaining. I still see no way around many APs for every room. I can run the cable, every room is not being utilized at the start. APs can be added as necessary. The important thing is the rooms are to be wired for them...copper and fiber. The more I read from contributors, the more I understand the problems and less I know how to proceed. Good thing I still have a two months before I have to buy hardware. Thank you for the insight.

 

[etorix]

I thought I read cat8 as not really capable of 40G speeds. In lab testing only, maybe?

Capable or not, it hardly matters as there's just no actual equipment out there which runs faster than 10G on copper. (Except for DACs of course.)
It's all fibre for speed. MMF OM3/OM4 fibre you could install now for 10G can do 25G as well, and I guess it will do 50G very soon. 40G/QSFP+ is a dead standard.
SM OS2 can do anything you'll ever want to throw at it, including the quad things (QSFP+, QSFP28,…), but modules are more expensive than MMF. MTP/MPO bundles are more of a thing for data centres.

 

[aefriot]

it's 1's and 0's, they either come through correctly or they don't. If the connection is marginal, then it might not present as a completely dead link, but you'll see a lot of framing errors/ CRC errors on the link layer (which are very obvious and easy to spot). In higher bandwidth protocols, 100g+ etc, you have extra things like FEC that come into play

armored singlemode fiber that will run 100gE without a sweat in house distances is like, pennies these days. no reason not to run it to be future proof for the next 50 years if you have the opportunity

ps 15 APs for barely 7k sqft is *insane*

yes, that many APs is a lot, but I have yet to test the conditions. I cannot trust that adobe will transmit well enough to maintain any respectable speeds. Still, from one rooms end to the other end will have to traverse 30-40 inches of adobe. I have not experienced walls that thick before. I have to set up a few test APs and temp network to test connections and speeds. Will likely post as things proceed.

 

[aefriot]

Capable or not, it hardly matters as there's just no actual equipment out there which runs faster than 10G on copper. (Except for DACs of course.)
It's all fibre for speed. MMF OM3/OM4 fibre you could install now for 10G can do 25G as well, and I guess it will do 50G very soon. 40G/QSFP+ is a dead standard.
SM OS2 can do anything you'll ever want to throw at it, including the quad things (QSFP+, QSFP28,…), but modules are more expensive than MMF. MTP/MPO bundles are more of a thing for data centres.

Thank you. Will keep that in mind. I have to put pen to paper to figure prices and if we dare try to push for speed even if equipment might expire before ten years and not yet be realized. We will replace the equipment likely before ten years.