Yes, it is a bit early to think in specific models, but it is good the know about options and experiences.
Now I know that x86_64 systems may be a great option to use with openWRT, better than raspberry.
But at this stage planning the cable runs would be the most important task, and where to put the conectors.
I will put coas in the salon and in the atic, and may be other place.
Is it possible to use the same tube for coax and a couple of ethernet cable or it is difficult to pass the cable?
Some points at the ceiling would be good too (for APs) and some in exterior (I plan to let there electric cables too in order to be able to extend it in the future, if I need to add lights or something like that).
May be I don't pass all the cable at first, but at least have enough tubes for it and in the appropiate places.
When I have a more detailed plan of the house I will ask for help about best places to put everything.
Well the laundry here is usually at the kitchen, just a washing machine.
I plan to put it in a room adosed to the house, near the kitchen. And it can be a good place to add telecomm near it.
It willbe well ventilated and place for hanging clothes will be in exterior.
My senior year in high school, NCAR offered a cradle connection for local high schools. We used rolls of paper punched with holes on a Teletype to prepare our code assignments. Later, at CU, I packed IBM punch cards to the Engineering Building that linked via Laser to the mainframes on East Campus. The laser did not do well during snow storms.
If I were building a home and had openwalls, I would run plastic conduit into well placed wall boxes. A fish tape and you would be ready for whatever transmission modality comes along. Also recommend browsing the smart home options including an uplink to your garage door opener, security cameras , etc.
Apologies for commenting again. I would strongly consider copying smart home builders in your own country, or region. There's no point running lots of copper line, and ending up behind the winning team, which will be highly dependable multi-gig wifi 6, wifi 6e, and wifi 7. Just sayin' . (Unless you have many microwave ovens that run continuously). I'm having great fun with these smart homes, & rarely run a piece of lan wire. Smart home gizmos are very cheap now, balancing prices between ebay and amazon shopping. Pax vobiscum. https://stevessmarthomeguide.com/home-networking-basics/
I don't trust wireless at all.
It is a good solution when there is no other solution and you need connection. But wireless is most of the times problematic and not reliable.
Wifi6 may be quick int the sort distance but not more reliable or noise prone in the long run.
And the IOT world (at least in security) is the worst place in my opinion to use wireless solutions: you can eliminate all the protections in one run with a simple frecuency inhibitor from "aliexpress".
When there are long cable runs and old instalations, wireless may be the best options.
Planning for the future and from scratch, wireless is not the best option at all, and I will try to keep it to the minimum and only if cable expenses go high or for needed points.
Maybe, maybe not.... in the traffic shaping/AQM department OpenWrt is actually pretty much in the top tier. After the wireguard/FreeBSD pfsense dramolett I would be thinking twice before committing to pfsense (but then I do not need any of their advanced features so that is easy for me to say). And there is OpenSense as "alternative" if you want something else besides OpenWrt.
OpenWRT is certainly capable but it's not ideal/does not always bring the full potential for all applications and/or hardware platforms. It's great for cheap APs but I'm not sold on maintenance especially if its off site but that's me valuing time.
Looking briefly at the RPI4 Community build thread I do start to wonder when you need to use userland software to fix kernel/driver limitations (like irqbalance) and other somewhat strange things.
@anon80727944
I think your ideas are good in general unless you're on a shoestring budget, pfsense and/or an OS that doesn't target rather limited embedded devices does offer a much better experience on somewhat powerful hardware and the upgrade/maintence path.
Just to be clear, OpenWrt runs fine on x86 as well and covers all of the above.
I mean, of course, it's always nice to have alternative implementations and also Linux isn't all that great in every aspect. But pfSense?
So, of course, shit happens. Everywhere. But I get the smell and I can see how that happened. I'd get nowhere near it.
Plus people might want things like MAP-T or MAP-E for DSlite in 2021.
If you want BSD, opnSense may be a good idea.
If you want something with commercial support and automated updates, but still open source, take a look at Turris devices.
edit: if you want high-performance hardware which runs OpenWrt, take a look at IEI Puzzle M90x.
OpenWrt runs on x86, it is however quite limited in terms of performance due to current hardware (arch) targets, (pre)set optimization and to some extent packaged software out of the box which has been been up for discussion before. Due to what OpenWrt mainly targets you still have to work around a lot of limitations so it's a stretch unless you plan to package/re-package quite a bit on your own but if that's the case you might just go for something that already covers most of it ootb and have more performance oriented optimization.
Definitely run conduit if you can. I don't think there's much point in going overboard with the initial cable drops: a couple of Cat6A pulls where it makes sense and coax where you think you'll actually use it. Leave a few pull strings in each conduit and it'll be a cinch to scale or upgrade when the time comes.
Under no circumstances would I commit to a wireless-only install. I have a small home (~150 m2) and even then I have never experienced anything close to full-gigabit speeds over WiFi, even with 802.11ax. Mesh networks might help there but with a wireless backhaul you're only compounding WiFi's latency issues. At the very least I'd want strategically-located cable drops to provide a wired backhaul for a mesh setup. Wireless has come a long way but it's hard to imagine it ever matching a wired connection for sheer predictability/dependability. Just my two cents.
All I can say is, Ring, Lennar, and DR Horton(1st largest home builder in U.S.) are building homes with inclusions like these, and maybe wire to three rooms, and it all just works! Why fall behind the times? https://ring.com/lennar
Because it is much simpler to not use installed cables than having to run cables post-hoc? WiFi is okay for what it is, but if it isn't it is good to have alternatives.... And the biggest issue I see is to make sure to install enough cable ducts/tubes at the right places (with acceptable turn radii and preferable pre-loaded with pull wires/string).
In my area, home builders have been using electricity for kitchen and not installing gas.
Vitro kitchens are the tendency. Everybody wants them, but not having gas conducts at home is not a better option than having them even if you do not use them.
They have saved a lot of money selling you what a modern home you have for having inductive or vitro kitchens and not having that old conducts at home. And they have saved a lot of money not installing them in the homes they sell.
Now electricity prices have reached the sky and all people who has no gas have no alternative (well gas has gone high too).
Home builders make what people asks for and what makes money for them.
I will lay as many cabling as I can afford and makes sense, and then put wifi AP only where it is really needed.
Wifi mesh is not a better alternative than cabling and switches, just a convenient option when you have the need of connection in disperse places or many mobile devices and you cannot get there using cable.
By the way, at leas with openWRT and the routers I have, wifi Fast Transtitions do not work with every devices, there are many android (<=10) devices that simple do not connect if you hav 802.11r (fast transitions) and 802.11w (management packets protection) activated at the same time and yet others (may be Android 11) that do not connect if you disable 802.11w so you have no way to have them all happy.
Have a read of that and think about what/when you are doing. What you are planning is more like an enterprise class deployment rather than a home affair.
Wired will ALWAYS beat out wifi. Hands down. Thus your plan for doing conduit and backhauls is fine.
Having your AP's hardwired will save issues in the long run. I've had to support mesh systems in buildings with no backhauls and its not fun.
Dedicated devices are going to be better long term. Having separate router, switch, wifi means later upgrades are less disruptive. "combined" type force compromises. If you are wiring rooms then a 24/48 port switch is going to be better to use as your backbone and then hand that off to a dedicated router. Ubiquiti or maybe something like the R4S.
I bought mine back in march and it was bit more costly than i wanted but combined with a ubiquiti wifi point? it completely replaced my BT Hub5 (cos its wifi was crap at going throu concrete walls). The R4S barely blips its cpu vs the 2core hub which would lag on updating AGH. I also added a small switch to deal with other wired runs. If I need more ports later I will just upgrade to a PoE multiport switch. (I don't have cameras now but may add in time)
The biggest issue is where to get one. For the 4 GB RAM - Combo with Metal Case :
Additionally 2.5gb and 5gb are still really niche right now. "Home" varients only really started appearing in 2020. Most are still enterprise class and thus cost. However if you start with 1Gb switch for now then a rip and replace to 2.5/5 or even 10gb when cheaper switches come out. It may even be cheaper to buy an ex enterprise 10gb switch as they migrate to fibre.
No shi*t, it is amazing actually how long these roughly a decade old MIPS router SoCs kept up... I hope the current crop of run of the mill routers will be useful as long.
Its only due to OpenWrt that i kept using it as long as i did. However I switched to Ubiquiti AC-Lite for wifi and its night and day in terms of performance. I could load OpenWrt onto it but I actually prefer Ubiquiti's firmware at present. They are stabilising V5 nicely now and i just run a docker with the controller on it. Its currently running on a pi elsewhere but with the R4S i could easily host it on the R4S router instead.
Thank you.
I don't want to end up with a mess of cabling either, there is a limit in how many cabling you cand lay and connect to one point.
I will try to think in advance and put at least conduits to some places I had not thought it would be needed (like in the ceiling).
If you can use one conduit for two ethernet or an ethernet and coax cabling, may be I can use just one for the TV for example, and one for a room with an ethernet point at the base level and then continue there another conduit to the ceiling (to put an AP there later if needed) and lay two cables in just one counduit to the base (and continue one of them to the ceiling from there) instead of having to use two counduits from the telecomm cabinet. It will save some conduits that would make difficult manage them in the cabinet.
Another cabling and conduit to the other wall would provide phone access too.
I will need to have the floor plan first in order to better think where to put each connection. I will ask here when I have it more detailed, in order to see if I have missed something.
I have now quite clear that I will need a managed switch with not so few ports (may be 16/24) and better with POe.
If it has 2.5 GBps (without going too high in price) I will use one of them, if not, 1GBps will do, and I will allway be able to add a second switch in the future for the most demanding devices (and let the old for cameras or other less demanding ones).
But as somebody said before, it is too early to thing in specific devices, as the house is not being built yet, it will take two or three years at least, and in that time all will change a lot.
I will add a main router in the cabinet and the ISP ONT or router (just acting as a bridge) and that router will be based on RPi or x86 and openWRT compatible (will see later which is the best and more economic model).
A roof antenna and an amplifier that will lay coax cable to at least the salon and atic (may be kitchen too).
The Digital TV receptor for distributing tv through ethernet may be will be put in the cabinet or in the salon, will see what is best later.
A couple of Ethernet connections in exterior for cameras and AP (one in the backwards and one in the fron at least) and lay ethernet cable and electricity to the main entry (probably a camera there) and garage.
But I see I will need a quite big rack in the cabinet, with 9 or so slots.
There is nothing stopping you just using a small 2.5/5gb switch as a master backbone and having multiple other switches below that. Eg. 24/48port 1gb big switch for most of rooms and a 8 port 1gb PoE switch for cameras / APs. Have both of them plugged in off the 2.5/5gb. That way you aren't paying over the odds for a full PoE switch when a basic one and regular nonPoE one is cheaper.
Yes probably that would be the most wise thing to do: a small 5 to 8 2.5 GBps new switch home prosumer grade, and another 1 GBps with PoE for connecting other devices like wifi AP, cameras.
So I will need space in the rack for at least:
Several power connectors (may be 6 to 8, one slot).
The backpanel connectors 1 slot.
The ISP and main router (1 slot).
two switches
The TV IP encoder and may be amplifier (1 slot).
So about 8 slots would be fine, But I will need to see how many cabling would the house need first.
My idea is to put it in a small room where the washing machine would be, the electricity governing switches, outsides of home with an entry thorugh kitchen or living room (in order to keep it not accessible from exterior).
May be another option is to not use a rack, and make some kind of built cabinet with a front door and where you can sreww the switches, with ventilation and electric outlets.
i edited his ars articles, so i approved in the dark ages, but have moved on to doing things 2022 through 2032 style, & use my own personal tech projections now, with much better results
