Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Adding a fiber link to my home network (stapelberg.ch)
263 points by secure on Aug 9, 2020 | hide | past | favorite | 209 comments


This is slightly tangential but if you are moving and fiber internet is an important consideration for you then check out this tool I made to check fiber connectivity in bulk:

https://gigahood.com/

It also checks within 500 ft of the address. You may be able to trench a fiber line out yourself at that distance.


If you're in Switzerland like the OP, the Swiss government has comprehensive maps of fiber coverage [0] as well as the speed you can attain [1] (there are more breakdowns, check Geocatalog > Population and Economy > Communication.

Even better, you can download the maps and load them into GIS software. I grab listings from real-estate portals then load them into QGIS, where I intersect them with zoning maps, fiber coverage, noise maps and public transit isochrones. I can then filter them at will.

[0]: https://map.geo.admin.ch/?topic=ech&lang=en&bgLayer=ch.swiss...

[1]: https://map.geo.admin.ch/?topic=ech&lang=en&bgLayer=ch.swiss...


Unfortunately, these maps are neither complete nor very accurate. For example, smaller ISPs and networks (like LiteXchange) are usually not included. :(


The best way I've find out if you're "on net" (have fiber going to your bldg or near it) is to befriend someone in the networking and telecom space who hoard KMZs files that map out what's installed (each telecom develops them and shares them internally and sometimes with customers). I know of folks who treat these like gold and carry them from job to job and have most of the US and other nations covered. These are the folks I call when I need an address lit up to see which options and which carrier is out there vs. contacting one who is going to try to work it thru them. Most metro areas if you're on net you can get dark fiber for a flat monthly fee. Construction and permitting will bring an additional cost and things are delayed quite a bit with covid.

I've known a few folks who've got their own dark (or point to point wireless) to colos in downtown areas where they and their friends will rent half a rack or less to drop in a router and buy cheap transit and become their own little ISP.


How does one find these people? They sound like an awesome resource to know.


For Australia, to check if FTTP is possible (without spending five figures to connect) look for the purple dots here:

http://nbnmtm.australiaeast.cloudapp.azure.com/nbnmtm.html

Unofficial, maintained for years by some awesome guy. Govt refuses to share.

If you are lucky enough to move house, right next to a purple dot you can save $many $thousands getting FTTP because your neighbour has already paid for it, and houses nearby that want FTTP will still be up for a few grand but not ten grand plus, like your neighbour already paid.

Sucks that Abbott the onion muncher lied and spent $10 billion more on copper (than fibre would have cost) plus took 4 years longer than promised to deliver slow copper MTM NBN because he blocked the just-begun original FTTP NBN rollout (with lackey Turncoat effectively spruiking that nobody will ever need more than 25Mbps)

Except they are politicians and we have the "The Game Of Mates"

FTTP would have devalued Murdoch's 25 year old Foxtel copper that we are partially using for slow NBN that drops out when it rains.

Would love to see Abbott and Turncoat fully ICACked for their plethora of dodgy favours to donor mates, detrimental to everyone (except them and their donors).

Don't even get me started on Pyne (the fixer) and his $200Bn diesel MTM submarines.


Yeah, not remotely accurate for me. I'm typing this on consumer/residental symmetric gigabit fiber (albeit a bit oversubscribed on the downlink side, but I still get hundreds of megabits both directions). Not listed in the results table.


Bummer. Admittedly, it is working off of incomplete and possibly old data. The data requires ISPs to self report.


Can you share more details on the data source for this?

Seems highly inaccurate for the somewhat rural addresses I’ve checked in Pennsylvania.


Sure thing. I am using the FCC form 477 data. Unfortunately it relies on ISPs self reporting their coverage data so it may not be fully up to date.


477 data is renowned for being wildly inaccurate. You can find whole articles about how terrible it is, never mind out of date.


Indeed. I wasn't able to verify more granular data doesn't exist but 477 data is at the census block level so I wouldn't expect this to provide anything different from https://broadbandmap.fcc.gov/


Yes. It is a more convenient way to search addresses in bulk. Also makes it easier to identify possible speeds within 500 feet. This is actually how I found my acreage homestead.


Is there any service that will monitor MLS/others and look for houses and apartments in a price range, within a commute distance (by train or car or bus), and with internet access?

I think I'd pay for a service that did this.


That wi be highly dependent on the MLS available in your area. There is no universal or standard MLS and the information collected can vary wildly between cities and counties.

Monitoring that checkbox if available would require access to an rss feed for the listings. Or potentially a realtor to set up an email alert (no contract would be required)


I would build that - but I have a feeling that the realty companies would prevent and/or fight you for scraping their data. Anyone have experience with that that can offer some info/advice?


I've scraped MLS for personal use to build a very similar tool (draw on a map, highlight new listings, able to hide listings), and had no technical issues at all, just a simple curl call. This was a few years ago, though.


This (and other aggregator websites) are consistently wrong for my address. Shows CenturyLink with 1000/1000 available, check CenturyLink's site, 15/10 max (lol).

Not sure where they pull their data from but it seems quite inaccurate. Max CLink has offered me for over a decade is 15mbps down.


Indeed. The data is only on a census block level and could have false positives and negatives. Unfortunately, the data is not more granular than the census block level.


Is that a fiber offering? Why would it be that slow?


ADSL. No fiber to my home.


Nice tool but at least in my area it’s wildly inaccurate.

(I know it’s not your fault)


Thanks, great tool! It gave me the same results that I found with a lot of manual research so it's definitely accurate. It's even more accurate than one of the ISP's own websites (I've confirmed multiple times with them that their website is wrong and your tool is showing the right info).


I'm currently looking into homesteading on acreage and your site is extremely valuable in that search. Thank you!


You are likely to be very disappointed if you rely on that data.

You may even be disappointed if a carrier promises service at a location, since some ISPs have been known to renege on offers at a later date.

The only way to be sure of service availability is to order service and have somebody turn up and actually install service.

The only thing the FCC data will tell you is that service ISN’T available at a certain location.


It certainly does require you to verify coverage. However, it is a good heuristic to see if you are in the right spot.

For my area in Denver it has been quite accurate. I recently found a 5 acre homestead property with fiber next door. I ended up trenching a line myself and I have gigabit speeds!


Maybe add a note on the page which regions/country this is intended for (assuming US only?)


Wish I had known about this months ago! Any more info you can share on data, plans, etc.?


I will have to check to see what the FCC provides. Are you interested in any data in particular?


Two kinds of data may be interesting:

1) where fiber may be run but not offered for residential service. E.g., future offerings.

2) where business fiber is offered. This would cover the dedicated internet (DIA) option.


This data is usually proprietary and/or behind a broker/vendor/carrier/paywall.

You can get some of it from Fiberlocator. Subscriptions to their service is $3,800 per year.


Group buy? I feel like finding 20 people who know how to use a VPN and are trustworthy would be feasible.


This is excellent! All the other ISP-finding websites are ad-ridden inaccurate trash.

Edit: yuge suggestion: make the results linkable. I wanted to share them with friends and had to make a screenshot, so you're missing out on visits.


Thank you for the suggestion!

I actually used the tool myself for my own house hunt. I disliked all the tools out there for the reason you mentioned. I will write an article about my journey in the future.


Doesn't seem to work with addresses outside of the US


Is this US only? I got "This address is invalid". I live in the UK at the moment.

Would be good to know a bit more about this tool. I'd love to get FTTP!


You might find https://www.broadbandchecker.btwholesale.com/ helpful. If you want to check an address and don’t know the phone number, use https://www.findmyaddress.co.uk/search to get the UPRN for the address and paste that into the BT tool.

I’d also love FTTP, but the state of affairs in the UK seems really disappointing at the moment. If you run a business and live in a rural area then the Government will contribute £3500 to the cost of an FTTP-on-Demand build. Unfortunately, Openreach’s engineering costs are so ludicrous that (in my experience) it’s still not a realistic prospect.


The UK lives in the dark ages regarding Internet. In Switzerland I already had 150mbit internet (in 2010+) while when I moved to London I couldn't even get more than 16mbit while living in Zone 1. Still these days I can't get that speed.


It doesn't help that the UK (and some other countries) have corrupted the definition of fiber by allowing pretty much everything to be called fiber.

DSL (running on the shitty, rusty phone lines) is allowed to be sold as fiber because fiber is involved at some point in the transmission (up to the DSLAM). By that standard, even 56k dial-up could be called fiber because there would be fiber somewhere in the datacenter on the other end.

HFC (Virgin Media cable) is also called fiber even though it's fiber to a certain point and then you're sharing a coax loop with dozens or even hundreds of people. It's even worse because the shared bandwidth can be saturated and the connection becomes unusable because of packet loss (could technically happen with DSL if the exchange's uplink saturates, but I've never seen it happen compared to cable networks).

There's also a naming convention of "superfast", "ultrafast", etc which creates more confusion. If you want to express speeds, how about you just quote the raw numbers?


Definitely agree. Only a small percentage of the general public here seem to understand the difference between FTTC (which is what we have) and FTTP/FTTH (which is what we need). Perhaps if they did then the Gov would have to work harder on FTTP delivery.


Yeah, if you live in a new building or further out of Central you can get things like g.network or Hyperoptic but even aa new apartment block build by Westminster council doesn't offer fiber.

Maybe this all will change when there is more Working From Home to be happening the coming years.


Weird, I get 100MB consistently not far from zone 1. You may live far from the exchange


Or, ironically, too close to the exchange.

https://www.choose.co.uk/guide/exchange-only-line-fibre-broa...


OP was getting that in 2010 elsewhere - it’s fairly amazing that London isn’t getting multiples of that now.


I live near Hyde Park ️


Thanks for the link. As I suspected, I can only get FTTC (80/20). FTTPoD is available but ... as you mentioned it's not exactly the best option.

Going by https://www.openreach.com/fibre-broadband - I see "Good news – we'll be upgrading your area to Full Fibre (Fibre to the Premises - FTTP) soon."

Anyone know how soon "soon" is?


Anecdotal but mine was about 6 months.


Can you elaborate on OR's engineering being ludicrous? I've found them to be quite helpful, and their costs to be roughly in line with the costs I would expect from other trades!


Well, the cost can be both reasonable compared to other market rates and yet ridiculous. I know someone who got a quote recently to connect their rural property to fibre, literally about 100 metres from the cabinet, like, not far at all. But because they had to dig through a road and lay the cable over a small stream they quoted about £60k for the job. That might be "in line with the market" but it's still a ridiculous amount, there's no way a normal homeowner can afford that.


Right but that's not openreach's fault, that's because they need to dig up a road and go under a stream.

Get a quote for a mains gas connection or an electrical connection and you'll find it in the same ballpark.


Well, so for electrical work I know that Northern Grid will do any reasonable upgrades for your network for free, just because that's part of what they exist to do. I asked for an upgrade to 100amp connection to my house, they did a survey and answered that they would need to dig out the whole cable from my house to the street and replace it with a new one. "Of course" - they said - "we would put everything back at is was once we're done". Cost? Absolutely free. Well, funded by the taxpayer I guess. I don't see why openreach shouldn't be the same - if your house isn't on the fibre network, they should connect you as that's literally their mission statement. I guess for really complicated cases the cost could be shared or something, but quoting people 50k for connecting a domestic property to fibre shouldn't be acceptable.


But it's not just 50k for a domestic property, it's 50k to dig up a road, and tunnel under a stream.

I wasn't aware that electrical connections were done for free - Citizensadvice [0] calls out that for gas if the connection is far enough away you'll have to pay for it. I've heard anecdotal costs of <£1000 for a straightforward connection that doesn't involve digging up a road.

[0] https://www.citizensadvice.org.uk/consumer/energy/energy-sup...


I have a home office situated 15-20m from an Openreach manhole. The building is less than 20 years old so there is an existing, unused, duct from the manhole directly to my home office. I live in an FTTC area so there is already fibre nearby.

The quote for getting fibre to the office was £9.5k + VAT (i.e. US$15k). And remember: this is for infrastructure which is long-lived, and which Openreach will have to install anyway, sometime over the next decade or less.


That's much more expensive than my experience, I wonder if there was another complication? I was quoted closer to 2k for installing to my building about 100m back from a cabinet.

They ended up installing it separately, and it looked like it took them about a week.


Unfortunately, I am using the FCC form 477 data which is only valid in the US. Do you know of a similar dateset in the UK?


Not accurate for me. Claims I should be able to get 1gig fiber from AT&T but AT&T site says 25mbs max


I checked this site for my own address and it's wrong, unfortuantely.


I've used some of the pre-terminated assemblies w/ pulling hardware from these guys: https://www.lanshack.com/PreterminatedAssemblies.aspx

The pricing is reasonable, especially compared to having a contractor come in and pull and terminate small, low-count runs, and the results have been very good for me.

Edit (now that I'm not on a phone): I like these assemblies because I get a "super power" feeling, being being able to cheaply add fiber runs for special cases. The most recent one was extending a single-mode campus fiber termination to a new closet within a building. Loss budget was very high (because the campus run was fairly short) so we did a simple "glass-to-glass" patch on one of these assemblies to the new closet. It was going to be a couple thousand dollars for a contractor to come in and do a proper fusion splice, but the Customer ended up spending $250 for favorable results.


As I wrote in another post, <1km fibre runs don't need fusion splicing. This is what I tell all people still insisting on using expensive multimode for local fibre setup for ease of splicing.

Multimode will start to struggle at distances at which even worst quality single mode setup will work just fine. I feel SMF has no sense economically these days.

Low power, single mode SFP transceivers are everywhere now, and in general, SMF setup is incomparably more future proof.


The photos show a lot of cables stuck to walls. I’m surprised people are ok with this.

Every house I have lived in has done it too and it’s the first thing I rip out.

I see contractors (usually ISP installers) doing this inside and outside houses too and it irritates me. UV, animals/insects, wind etc gets at them and water goes in the holes. The fixings that hold the cable rust/degrade and fall out.

Doing it right costs more and takes longer but I’m unsure if the cost is greater when you look back 5-10 years.

Maybe doing this is slightly more repairable if the place is a rental?


You may underestimate the time and effort that can be involved in in-wall cabling in a retrofit situation... like most in my area I live in a house that is coming up on 100 years old, while I do my best to run all cables through walls there are situations where I have to run cables in conduits on the exterior as the only way to do otherwise would be to resheetrock most of the ceiling. There's just a cost-benefit analysis here, and most contractors aren't willing to put in the time (because the customer doesn't want to pay for the time!) to open up the walls to get access to run cables through them. Keep in mind that when you consider demolition, hanging sheetrock, texture, prime, paint, and clean, any cabling situation where walls need to be opened is probably going to be billed as a full day, and local building/licensing codes may not allow a low-voltage technician to perform that work, requiring a builder to be subcontracted.

There are tricks like flexible installer bits that allow for short runs through-wall without removing the wallboard but you're going to need an access hole every couple of studs at least... and at that point you're setting yourself up for so much patching it might be faster to just open up the wall entirely. You can also use all kinds of wire-pulling tools, I have a 30' telescoping fiberglass pole that I have used to run cables through inaccessible attics (flat and shallow roofs common here), but once again, you get into situations where running the cable for a single surveillance camera is an 8-hour job! I put up with this for my own house but clients aren't so happy about it...

And all of this is assuming wood-frame walls which are the norm for interior walls but exterior in many areas can be brick, adobe, etc., and blown-in insulation is great for heating bills but can make cable runs in exterior walls a huge headache.

At the same time, cables run "cable-installer style" (stapled to gables and run down exterior to straight-through holes) regularly last for over a decade and are very easy to repair when there are problems and modify when needs change. It's hard to blame them.


It cost me $3000 to have a professional run 7 CAT6 cables to every room throughout my 80 yo house, plus one to the modem, along with wiring proper earthed power into another four locations or so. The man was not claustrophobic. Plus less than $500 for another contractor to fill in and paint the sheetrock that was required in storage rooms. The cables meet in the basement in a server rack.

It was well worth it to me. I have teenagers and everyone has an Eero in their room, plus hardwire for the home theater and office.


That’s a lot of money. I’ve run about that many and while the the material cost was relatively low, but it has been days of work and a lot of frustration and mess. I think you picked wisely.


> You may underestimate the time and effort that can be involved in in-wall cabling in a retrofit situation.

I totally hear you. I’ve done a lot myself, most recently on Saturday.

It’s hours and hours per run. I’ve used many weekends doing this and it sounds like you have too (and it seems you have considerably more sophisticated tools!), but the alternative is looking at cables that run along the walls.

I ran into a new one recently - the nogs (dwangs to some) being angled to give a herringbone pattern. A complete nightmare for retrofitting.


Agreed here. I'd love to run cable to access points on my 1st floor ceiling, but the work required to do that (getting it up from the basement into the ceiling and through the joists) is just too much work. I've decided just to run the cable on the ceiling.

If I ever build or renovate a house, I'll be running it like that though.


In new construction, consider running "smurf tube" or other conduit between structured wiring enclosures. You'll be able to pull through more cables in the future reasonably easily, especially if you leave pulling line for yourself. In theory you can also get network appliances and etc. that mounts nicely in the structured wiring boxes although honestly the selection/pricing is kind of lame on those. This is starting to be more common in new construction, esp. multi-family.


You also have to consider the author is located in Switzerland, which has a much different construction style when compared to, say, California.

In order to actually lay wiring underneath a masonry wall, you need to rent/own a wall chaser, be aware of where existing cable runs are, and be prepared for a lot of work: prepare the room for dirty work (cover the floor and furniture, isolate the affected rooms from the rest of the house), do the actual cutting work, lay the conduit, re-plaster the entire run, re-paint it (sometimes re-paint the entire wall if you don't have a good color match), clean up. You're basically looking at an entire day of work if you're doing this solo, plus prep. And it's pretty much never doable in any rental.


I have been doing this in my house. Takes much longer than a day. For example I installed 2 Dolby atmos systems, with pipes for HDMI, network, power as well as 9 speakers - took many days in total.


are hdmi cable durable? i have too many bad experience with faulty hdmi cables so i never think of doing this.


Having run a few to ultrasound machines, never have the plug coming out the wall, have a female plug on a wall plate at each end. That way a ruined plug can be fixed and a new wall plate added without redoing the drop.


In residential places in Europe I‘ve seen this mostly in old buildings where the walls are stone/brick underneath, so I am assuming it would be hard to put the cables inside walls?

Also, this was always inside, not on the outside walls, plus the cables were laid in tubes fixed to the wall and painted over, so well enough protected, probably.

Did you see issues with decay with this sort of installation even inside an appartment?


I've seen damaged cables from kids, cats and dogs. Ive also seen water damage inside, thought that was in the termination so could be fixed. My main objection is that I don't want to see them, with damage coming second.


Depends on the walls. At my parent's, the house was built ~60 years ago with solid brick walls, no space to run fiber or copper cables.


Is there a basement/crawl space? I have cut a narrow slot in the wall and then drilled down to the basement/crawl space for this. Messy, needs plastering and repainting but avoids exposed wiring.

I have no idea what you do if there is no basement. Go behind skirting boards?


You wall chase, then run conduit inside the wall, then repair.


My internal walls are solid brick. In the last 3 years, I've moved my router twice (moved from DSL to coax and to FTTP),and I've ended up running Ethernet cabling around one room.

Fixing the walls after cable runs is nasty. Underfloor is possible but usually involves pulling up a floorboard or two which isn't ideal...


A tip for surface cabling without major digging for houses with baseboard is to use a router to make a channel in the backside of the baseboard and just run in that. I've seen electricians in the US do this for outlet wiring, in addition to phone/data.


What’s the right way to do it?



If you are building new or renovating, run the cable pipes before the walls are in place. If the walls are already there its much trickier, but still doable with a long bendable drill and magnets, and you might end up with the switch on the attic as its easier to drop down.


I find going up from the basement to be easiest. Dirt rocks and spider webs beats heat, grime and ceiling creatures.

A lack of space above or below and solid brick walls are going to be seriously limiting. Possibly the only upside to New Zealand’s average house quality is that there is space everywhere and most things are hollow.

> you might end up with the switch on the attic

UniFi makes stuff I like and this situation sounds like a great use for their ‘Flex’ switch which is managed and POE powered.


I find going up from the basement to be easiest. Dirt rocks and spider webs beats heat, grime and ceiling creatures.

I think it depends on what creatures you have in your area.

In my location, I'll take ceiling creatures (bats, owls) over basement creatures (multiple poisonous arachnids, poisonous snakes, poisonous lizards, etc...)


Can confirm, DIY fiber is not too difficult to figure out yourself. For most uses outside of an ISP you don't need the super-low loss that you get with a fusion splicer.

When I needed to wire up a campground and some cabins I studied a couple of the Fiber Optic Association textbooks and I recommend them as informative and easy to read. Also it is pretty easy to spend $50-100 over and over again on "one more tool" to save time and effort.

On a random tangent, DIY fiber topics often makes me think back to a book about Kevin Mitnick (Takedown I think) where the author describes how he has fiber running around his house for various reasons. That was pre-1995 which was when the book was published; I imagine it was much more difficult and expensive back then.


I honestly found it no harder than anything else - I just did a 300m run to an outbuilding, as 802.11g just wasn’t cutting it, and the project basically compromised two media converters, a reel of armoured fibre, and a termination kit - took all of an afternoon to do.

I guess it’s just overkill in most scenarios - but for long runs, you can’t beat it. I’ve also got a 200m cat6 PoE run with a repeater/voltage booster halfway down, and it does duplexed gigabit just fine - which even further limits the scenarios in which fibre is the better option. Even with the repeater it was cheaper than the fibre alone, never mind the media converters.


My training in the US Army included learning to splice fiber optic cable by hand, including polishing the end of the fiber itself! That was in 1993.


This bend-insensitive fiber the article mentions, combined with the narrower diameter than Cat-6, might make it more practical to run across the tops of mopboards in my rented apartment.

WiFi speeds are usually OK, but the apartments density is so high here (and new `xfinitywifi` APs are often popping up on my channel), that I'm almost ready to move back to cabled for most purposes.


If you want thinner than normal Cat6 but can handle a bit more than 0.9mm, there is also thin Cat6 available: https://www.monoprice.com/product?p_id=34213

It's 2.6mm diameter, and I had success running it between baseboards and carpet.


Thanks! I'll have to try at least one run of this thin Cat6.


I switched to cable when I moved to an apartment. Channel 11 was becoming unusable right around 8:00 every morning when all the WFH people got online.

My apartment has a nice patch panel though so I don't have to actually run cables everywhere.


Were you stuck with 2.4GHz for a reason? I can't imagine how bad that would be in an apartment.


Depends on the apartment. In some, it's not as congested as it used to be because so many people have moved to 5GHz.

It's still amazing to me, though, that so much new gear is still 2.4GHz only. I bought a brand-new, top-of-the-line air filter earlier this year, and even at close to $500, 2.4GHz is the only option.


....why would an air filter have WiFi in the first place?


You could ask the same thing about light bulbs, but they do.


You often want to control brightness and possibly color on light bulbs, and it's good if that can be integrated with other systems so you can have central control and automation. Wifi may not be the best way to connect, but it's reasonable.

With an air filter, do you ever want to control it? Naively I'd think that you'd want to leave it on unless you're going to be gone for a week or more. And even in that case it's probably easier to flip the switch.

Would it tell you when it needs a filter change? Would that actually work better than a calendar?


While I somewhat agree with you, it's possible to do some pretty advanced/nifty things with advanced automation platforms like Home assistant. There are prebuilt extensions for things like air quality index, so in theory, you could set up a rule to run your air filters when the exterior air quality is above a certain threshold of dirty. The same could be done for indoor air quality if you had a local sensor monitoring it (can be done pretty cheaply with a Pi but I've seen off the shelf products as well)


There's also powerline, which works very well for me and all you need are power sockets to bridge the distance between your desk and the router.


Powerline can have major problems, depending on the quality of the electric wiring. In addition to that latencies are bad.


I once had powerline networking which somehow ended up joining my next door neighbour’s powerline network. That was a bizarre evening of trying to work out why my connection was sometimes fine, while other times unusably slow - turns out I was occasionally getting DHCP responses from next door’s router, and unsurprisingly the available bandwidth across RF interference between two power sockets back to back on opposite sides of a wall leaves a lot to be desired.


that's funny. how did you find out?


Thankfully the neighbour did freelance networking, so the domain being handed out to DHCP clients on their network matched the business name. It turned into quite a friendly relationship with them, I hadn't previously known that they were in the tech world as well.


depending on the electrical wiring, latencies may not be that bad.

Ethernet is usually <1ms.... powerline adapter i use is about 2ms.... wifi to the same router, is 40ms. 2ms is much more manageable for me than 40ms


40ms wifi is surprisingly bad. I'm getting 2ms +- 1ms to my router (macbook pro -> ubiquiti AP -> switch -> router) over 5GHz wifi.

My WFH setup is on wifi and I've never had any throughput/latency/reliability issues that made me want to run a cable.


Yeah can confirm for the same use case though I use Ubiquiti Edge Router X SFPs off a Ubiquiti Edge Router 12P so I get more ports.

Of course at this point we're close to Wi-Fi 6E hardware and I probably wouldn't have done all of this at this point.


A quick search shows that single strand fiber is available with a white outer color. That would almost visually disappear in your mopboard example.


You can get fiber even as a translucent version, that is almost “invisible”.


They are so thin I covered mine with the wallpaper


What's oddly left out in the article is the detail of whether or not he already had a network switch with ports supporting more than GbE - the price for these things start at around 250-300 CHF which would double his suggested total cost - or if he's just linking two computers together. Personally I would just go with 2.5GBASE Ethernet and compact-diameter cabling. 2.5GBASE PCIe cards cost around $15 a pop.


QNAP has some fairly cheap routers that have a couple 10GbE ports, like this one: https://www.qnap.com/en-us/product/qsw-308-1c

I paid $200 for mine. Not cost-effective if you have more than a couple 10GbE-capable devices, but great for hooking up a fast NAS to a couple primary workstations and leaving the rest of the network 1GbE.


Wow, I've spent dozens of hours looking for a suitable switch at home and QNAP completely passed me by, thanks for the link. It looks really nice.

I already ordered a used Brocade/Ruckus ICX for $200 though. They have 12/24/48 RJ45 port models with an additional 2-8 SFP+ or higher and come in PoE versions.

These are prohibitively expensive new but can be found at significant discount (10-20% of retail price) on e.g. eBay. This goes for Cisco as well, though I haven't researched looked into that. When buying enterprise gear, many brands (these two included) require particular licenses to unlock capabilities on the switch, so one should verify they have/get the appropriate license as well.

Netgear's SOHO/Enterprise range, Mikrotik and Ubiquiti are also very popular.


Thanks for the name-drop, the brand was new to me, though it has just regular GbE ports unless I'm seeing the speeds wrong. 2.5GBASE/5GBASE is something that would be interesting as it works on regular Cat 5e/6 and 8P8C.


I think you're seeing the speeds wrong :P Not all the ports are the same speed.

> ... providing three 10GbE SFP+ ports (with one 10GbE SFP+/RJ45 combo port) and eight Gigabit ports


No I'm sure I'm reading it right: the 8 ports on the side are the interesting ones which you can connect your regular wired devices to, as 99.99% of wired consumer networking is on 8P8C Ethernet, not SFP+.


There are 8P8C copper SFP(+) modules so that isn't really a barrier.

Anyway, what you're looking for is something more like:

https://www.qnap.com/en-us/product/qsw-1105-5t


I think it’s a different usage you’re thinking of. Presumably you’d like 10Gb everywhere, while the article and OP are presumably aiming to link a cor switch or high usage device (NAS) at 10Gb while everything else remains 1Gb.


I'd like anything over 1Gbit/s within the LAN, between all of the LAN clients. 2.5G is a very affordable drop-in compatible upgrade - plain Ethernet cables and PCIe cards - but I've been having problems finding a decent SoHo switch with 2.5G+ ports. The QSW-1105-5T recommended in this thread seems like a good solution.


I suspect one reason he's doing it is for the super-slim cabling.

You can get super-slim ethernet cables, but they have really fine wiring and don't go very far. But with fiber you can go vast distances with no problem. You can even get GBICs that will take you 10km.

I've been considering it for... well, ok, no reason.

I just happen to have two openwrt routers that have an SFP port (mikrotik).

They've come out with some new ones that are all SFP port models, for not much money though the 10gbe ones are a bit more.

https://amzn.com/B07LFKGP1L

mikrotik 1x gbe ethernet, 4x SFP 10gbe, $138


That's the first time I've heard of success with Microtik and OpenWRT - do you have more details?


Openwrt supports quite a few mikrotik devices:

https://openwrt.org/toh/views/toh_available_16128?dataflt%5B...


No, all other devices are gigabit ethernet switches only.

I’m linking two switches together with this fiber, and the rest is copper ethernet cables in these switches and other switches :)

Behind each switch are 4+ devices, so individual network cards won’t help here.


You have 4+ networked devices in your kitchen, which require more than one Gbit/s network transfers?!


Yes, I have 3 devices that require fast networking: I run the primary network storage replica, a build server (only powered up on demand) and a mirror server.

A Sonos speaker is also connected to the network, but that’s obviously neither important nor significant in terms of traffic.

In this article, everything is 1 Gbit/s, including the fiber. I’ll talk more about the 10 Gbit/s aspect in the next article.


Also check out https://mikrotik.com/product/crs305_1g_4s_in which is another popular and among the affordable home 10gbe switches.


I’m looking at using fiber and media converters to connect 7 houses and 2 entrances in a security estate. Main reason is to get IP cameras (and shared access to them) at various points but reduce the impact a lightning strike would have. We have a lot of lightning in this area and I’d be bleak if one strike wiped out multiple houses worth of network and gear.

If anyone has done anything like this, advice welcome.


1) Never never never never use copper ethernet between buildings. If lightning hits a tree or anything even remotely close to a buried cable you could be easily looking at frying every single device on the entire network in all buildings -- not even kidding. This happened to a popular tech youtuber TWICE before he switched to fiber.

2) If you want to know what went wrong and where use switches with SFP slots instead of media converters. This also makes fixing problem cheaper and gives you flexibility in choosing optics. Even if you do go with media converters, use ones with modules.

3) Build intra-building topology with some form of path redundancy. Since the cost of doing anything like this is so heavily weighted in building the physical links, you may find that you can build a redundant ring topology (only one extra link) at a 10+gbps at a lower cost than a non-redundant hub and spoke topology at 1gbps if the latter requires longer runs. Well maintained UPS's at each node are more important in rings, so think about this too.


Funny. Your first sentence had me wondering whether you're south african. Then you followed up with lighting strikes :-) Agreed, I think your best option is fibre. A few Mikrotik's (reasonably priced) and SFP modules will work perfect for you. I think just getting your own splicing kit will make this easier for you and allow some experimentation.


Don't forget that outdoor cable oftentimes has a metal strength member in it so it can be run from poles. The underground direct-bury kind of cable has metal armour for rodent-proofing and protection when the ground settles. Just because it's fiber cable doesn't mean you can forget about lightning protection and proper grounding. If the airwaves aren't too congested wireless might be a better option here.


You can always blow metal-free fiber in pipes. I'd use PE pressure pipe of suitable size, pull a string into it with a vacuum cleaner and pull fiber in... Hopefully the fiber survives.


Joburg :)


I ran fiber between my house and detached garage (which also includes my office). Installed some large conduit so the preterminated fiber could be pulled through, ran it just like any other cable (trying not to bend it too much), and installed SFP modules in two Unifi switches. I'd never done anything with fiber before, and it was incredibly easy.

I opted to install twelve strands of single-mode fiber. It's insanely overkill (I'm only using two strands right now), but the cost difference was minimal.


With something as complicated as 9 separate locations, I'd definitely avoid unmanaged media converters. Managed ones will be able to tell you how well, or how poorly, your fiber links are doing.


Metallic ethernet is explicitly not designed to be used between separate buildings so you really should not use it for that application. The physical layer can survive surprising overvoltage conditions, but it is insufficient for lightning strikes and there is no sane way to add surge protection at the building entry as would be done for telco outside plant copper wiring.


Skipping the field-terminated bit, I just grabbed a hunk of preterminated fiber and some dumb media converters. It creates a dielectric gap in my network, so lightning and surges coming in on the cable modem, can't damage the rest of the network. (I used to fry a wifi router about twice a year during storms, and got sick of replacing them...)


Wouldn't you still blow one of the media converters twice a year?


Probably not because the inputs were previously blown due to either ground shift voltage or EM induced voltage in the long Cat5 run itself. Neither occur with dielectric cable.


Bingo. Now the whole cluster associated with the cable modem can rise and fall with it, because there's nothing conductive tying it to a node at a different voltage.


Blown media converter is better than other equipment in your setup getting murdered.


Armored fiber is another useful tool for the home or small business user. Much less fragile which can be a big help depending on your application

https://community.fs.com/blog/what-is-armored-fiber-optic-ca...


I'd recommend using managed devices as opposed to using dumb media converters. If there are issues, you have no insight into the issues with unmanaged devices.


To expand on this - instead of using a media converter, use a small managed switch.

A 'dumb' media converter is a little bit cheaper but there's no way of knowing if the failure is due to the cable, the media converter on either end, or what. Without physically inspecting it, that is.

All fine for something inside your house but not something where you have a remote location.


Better yet, if you're connecting two devices that have PCI ports -- an eBay Mellanox ConnectX3 is great. Solarflare also works, but Mellanox has had better driver support in my experience (though we'll see now that they are owned by Nvidia).


I just got fs.com to make up one for to the right length. Worked nicely


A much cheaper set:

$13 single mode 10G SFP transceiver set [quite uncertain what they are now... maybe something not quite ethernet compatible]: https://www.aliexpress.com/item/33040961103.html

2x $20 DLink 10G SFP [a much more certainly standard compatible]: https://www.aliexpress.com/item/32847226618.html

$50 1km G657A2 bend resistant single mode cable: https://www.aliexpress.com/item/33045053714.html

$20 20x gel LC connectors: https://www.aliexpress.com/item/32810983763.html

$15 20x gel splices: https://www.aliexpress.com/item/4000183799694.html


That’s not a bidi SFP. The attenuators aren’t needed either.

Then there’s the fact you are ordering from Aliexpress, rather than Fiberstore. Fiberstore has multiple locations and warehouses, a returns policy, a level of customer service and no shipping charges.


Yes, it's not a BiDi, and I think they messed up with the text. I picked the attenuators for the power of proper 10G-LR, so yeah, they are probably not needed for the short range transceivers.


LR optics don’t require attenuators.

OP didn’t mess up the text. They clearly wanted bidi optics as they only ran one strand.


Can you identify what those 1.4km transceivers are? https://www.aliexpress.com/wholesale?SearchText=1.4km+10g

Are they even remotely standard compatible?


1.4 km is a curious distance. Perhaps they are optics that didn’t make the 2 km cut and are sold at a discount?

I wouldn’t worry too much about the standards. There are lots of optics that don’t stick to the standards, but go above and beyond.

I’m more interested in the use case at hand and if the specs will support the use case.


I saw a lot of singlemode wired datacentres in China, worked on them, but never bothered to look what they use for hardware. I wonder if this is the proprietary semi-standard that big Chinese telcos push for top of the rack hardware.

I once heard that Huawei was trying to do singlemode with VCSEL laser, but the descriptions clearly identify those as using 1310nm DFB http://www.6comgiga.com/downloadRepository/826d97f5-c91a-454... . It is not a 2km multimode


I mean, the item on the aliexpress. I always thought that 10G LR has quite a big minimum length.

Edit: in practice, only 2m minimum length with normal LR transceivers actually.


Maybe watching Comcast employees use a microscope when fusing individual fibers together when installing their service has made me feel like sticking my own connectors on fiber is a bad idea.

The thin diameter is a nice feature, but the ease of breaking them would keep me from installing them anywhere copper wiring provides adequate performance.


People have to be careful when splicing fiber when every dB of loss is important, but modern fusion splicers make it pretty easy.

For just running between houses, I've seen links working over fiber that was cleaved with linesmens pliers before a crimp on connector was added.


So I searched up "fusion splicers" and apparently there are machines that automatically align the fiber and fuses them together, with only ~0.02db of loss. Does a human doing it manually with a microscope get significantly better results?


Fusion splicers are very expensive, under $10k is the cheap end. ISPs own them of course for field work but home installs are often done by small independent contractors (one-person companies are common) without the budget. The situation has improved a lot over the last few years, but historically fusion splicing was such a delicate process with sensitive equipment that ISP crews would pull around an air-conditioned trailer just to house the splicer and its operator. Actually these trailers are still the norm, but the machines have also come down in price and up in flexibility so they aren't so necessary any more (but I'm sure it still really helps productivity to have your tools on hand and not be dripping sweat on the workspace if you do this all day...).

But in any case I find it more likely the installer was using an inspection probe/microscope to verify a field assembly connector than actually splicing. Most field assembly methods don't require a microscope but it's common to use a specialized type of microscope to check the correctness and quality of the face of the connector before use, to avoid future callbacks, which said independent contractor will be penalized for.


Fusion splicers are cheap. You can get a brand new fusion splicer for under a grand.

Fusion splicing hasn’t been a delicate process for at least twenty years and trailers are used only because they provide better working conditions.


Expensive if you need to be competitive with higher volume.

Old fusion splicers can be inexpensive. Tradeoff is they take longer to to do alignment and are much heavier than modern splicers.

I got a FSM-30S for sub $3k on eBay years ago. $350 for calibration. I see listings now for under $500. Paid for itself very quickly vs. hiring out as a little ISP. Wasn't hard to learn from YouTube videos and material from Fiber Optic Association.


FYI, brand new fusion splicers can be had for under a grand.


Per-splice costs of crimps are higher, but startup cost can be lower. Tools for crimp are generally $1k or so and splicers used to be $10k+ (though now you can find new ones for sub 3k and used for lower than that). A mechanical splice can cost $10 in material while a fusion splice might cost $0.05. Consumables for a fusion splice is just a metal rod for heat shrink tube that goes over the fiber, then a metal rod next to it, and another heat shrink tube to hold it all together.

I haven't seen someone use a microscope to inspect a splice, but it is common to inspect the connector face because dust can significantly impact the link.

Fusion splicers are amazing pieces of technology. I used used to use one from the late 90s that would identify the core (super small, usually what you see as "fiber" is the cladding around the core), position them to face each other, warn if the cleave wasn't parallel, arc weld it together, detect any bubbles or imperfections in the weld and alert.

Decent video of the splice process on the splicer I used to use (though missing fiber prep): https://www.youtube.com/watch?v=z4eRhF2oaqA


Crimp tools from brand name vendors are wildly overpriced. You can get a mechanical splice kit for very little money from one of the online fiber stores.

In fact, for less than a grand you can get a brand new core alignment fusion splicer.

A mechanical splice isn’t $10, it’s more like $1 in bulk.

As for consumables, fusion splicers also need new electrodes and calibration from time to time, but these are minor expenses.


Nope, a human with or without a microscope isn’t going to do it better than a fully automated fusion splicer.


I don't think people people should even bother with splicing in under 1km fibre runs if mechanical gel splice is 1000 times faster.

A bigger issue for single mode on short runs if ensuring transmitters are not too powerful. This is why you often have to intentionally attenuate single mode runs.


Solid glass has better mechanical characteristics than mechanical splices and a longer service lifetime.

No short haul optics need attenuators.


Yep, I just learned that modern standard 10G-LR does not need attenuators even for really short runs.


It's actually rather difficult to break cladded G.657 fiber, and it's much cheaper than copper. Now, at the splice point where you must strip the cladding you are correct it is very fragile. At that point you heat shrink on a splint with a steel rod in it which protects it (and typically place that in a splice case).


For sure, if copper works, it’s more practical and slightly cheaper for most situations :)

Regarding the connectors, check out https://www.fs.com/de-en/products/35165.html for details. In particular, they come with a pre-installed and polished fiber, that then only needs to be matched to your fiber using an already-installed matching gel.

It seems like a pretty well thought-out system to me thus far.


FS is one of my favorite websites for design, service, and quality of product. I prefer a cheap fusion splicer ($1500) as I find they work as well as a Sumimoto ($15k), but for one off home projects a mechanical splice with a FS pigtail works great. A 10km SFP ran in home (what, 1km tops?) can handle a fair amount of db loss.


Exactly! For my 10 meter run, it’s easy to make things work, even when they look like they shouldn’t: https://twitter.com/zekjur/status/1286570347337330689/photo/...


Haha that picture both hurts and seems normal. I've buried a splice in a Frozen themed tupperware that's still going 4 years later. Take that $120 underground splice closures.


Fusion splicers haven’t used microscopes for at least twenty years. In fact, I bet it was far longer ago than that.

Are you sure it was a microscope or are you really old skool?


I've run fiber from my basement to my second floor rack (two cables). It's really cool. Right now its 1G, but I can upgrade it to 10G whenever I feel like paying for new switches.


Back in university, my old roommate did networking at a car dealership and brought home old 10Mb fiber connectors. Yes, 10Mb .. designed for range, but certainly not speed.


So what? Running the actual fiber to my shop was far less interesting than digging the trench with the backhoe, running conduit, and backfilling the trench.


Does anyone actually know how these transceivers work? Is it software or hardware? They can pass an amazing amount of traffic and I’ve never had them crash.


I wish home grade routers would start using SFP already. Very few of them do. Why is that?


> home grade routers

That's your answer. The average home user just wants to plug in a device, type in the wifi code, and never think about it again. Only tech nerds want fiber type connectivity. Putting SFP into home devices would see that SFP sitting unused 99.9999% of the time. Why incur the expense of adding it?


In fact only tech nerds want wired connectivity at all. Ethernet ports have pretty much disappeared from all laptops.


A direct ethernet port is more about wired connectivity on the go, which is much more of a niche than wired connectivity in general.

At work/home you can have a docking station. Or a $20 USB hub that has a builtin ethernet adapter. Ethernet for fixed use is easy.

And it's legitimately difficult to fit an ethernet port on a typical laptop (or fragile, if you make it fold).


Also to the point that to many many people, "Wi-Fi" has become a synonym for "Internet connection"


Well, that's until they are hit with higher latency than they want. Wired connection quickly becomes a preference in such case.


GP is more talking about that the word "WiFi" now has the meaning "Internet connection", not necessarily wireless.


Good ones always come with Ethernet port. I wouldn't even consider buying one without it.


I have a MacBook, which don't come with ethernet for the last 4 years. But I do want ethernet. So I simply dock my laptop, and the docking station of course has ethernet.

Of course I do understand that if you're a network engineer, you don't want to screw around with docking stations and dongles and such.


I don't consider Apple laptops good to be honest. Things should be simple, but not simpler than simple.

That said, if some device is simply too thin to accommodate a network port, I don't mind using an adapter through USB-C / upcoming USB4 and etc.


> I have a MacBook, which don't come with ethernet for the last 4 years.

Exactly: All good laptops come with Ethernet ports.


I thought my new xps 13 was good. Guess not


If it's too thin for it, it's fine - it's a valid excuse. But if it's not and it's just a trend, then no, it's not good.


How is regular Ethernet connector better for it? SFP can as well be same level of easiness, if devices actually supported it. So why not?


Cost immediately comes to my mind. Given that gigabit service is the gold standard of residential Internet, it is not clear what would be gained for the cost of SFP ports and modules. It is also unlikely that we will see residential service reach 10 gigabit speeds, as most consumer devices would never be able to take advantage of such a thing, and even at 10 gigabit it is probably cheaper to build everything as SoC rather than have separate modules. The only reason I can think of to bother with SFP for residential connections would be GPON, but according to a Verizon tech I asked about this it is far cheaper for them to just use their ONT (which also supports their triple play package), and I was the only person he had met that even had a router with an SFP port.

Until you are looking at speeds above 10 gigabit or need to run particularly long cables (beyond 100m, which would almost never be relevant to most consumers) there is no particular need for SFP, and the added cost and complexity is beyond what most consumers want to deal with. Given that most consumer devices cannot come close to saturating a 10 gigabit link and people mostly use wireless links anyway, there is no real point in putting SFP ports in consumer devices.


I've read somewhere that cost of SFP dropped quite a lot over the years. So it's not about the cost anymore. It's just inertia.


It is definitely about cost. An SFP module for 1 gigabit copper ethernet is $20-30, and a typical consumer router has 5 ethernet ports. For $100-150 you can buy a consumer router with 5 ethernet ports, a wireless card, and even USB ports (to e.g. host a storage server or connect to a printer).


You don't even need to have many SFPs in a router. Just a few is good start.


Consumers would see it as a downgrade: "Wait, I have to buy something else to connect this to my PC?"


Not really a downgrade it if offers 10 Gbps easily.


Is there 10G residential service where you live? I am not sure I have heard of anything above 2G, and even that was just a future plan last I checked.


Even transferring things between home computers at 10 Gbps is useful. It's not always about external connection.


Netgear, Mikrotik and Ubiquiti have some that at least are in the price-range of "home".


Too bad there isn't something like Linksys with open drivers that also has SFP.


I have symmetrical gig fiber service, and this says it's not possible.


I love this:

> I had only ever seen 2mm fiber cables before, and the 0.9mm cables are incredibly light, flexible and thin! Even pasta is typically thicker:

> (Image of some disorderly 0.9mm thin fiber cables)

> Preparing a delicious pot of glass noodles ;)


i was recently in market for a tv. surprisingly, most of it doesn't have Gbps ethernet but just 100mbps.

i ended up using wifi


any other reads like this? maybe a bigger one, like connecting a whole building or connecting to an ISP?


This is a very unsubtle ad for the website it links/mentions by name about a dozen times.


I’m not sure I agree - I’ve always been a fan of FS for fibre stuff, at least in europe - their prices and support are both excellent !


Yeah I got the same feeling that this is an ad, or at least in part sponsored by fs.


FS has not paid me to write this, I’m just a happy customer. I paid for all the gear with my own money.


is that 10gbit or 10Gbyte?


10gig ethernet is bits.

100gbit ethernet does exist, but its not overly cheap. That will get you to about 10gigabytes a second. more or less.


Network speeds are typically measured in bits.


10 gbit/s, as is common in networking.


a lower-case b (10Gbps) should indicate bits while uppercase (10GBps) would indicate bytes. This is the correct notation but you definitely see it done incorrectly fairly often. It's usually easy to tell though as giving network media speeds in GBps would be very unusual.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: