That is in London though - probably not where the problems are on a national scale
That is in London though - probably not where the problems are on a national scale
They have symmetric gigabit fibre in all of the villages, for €20 a month.
In my apartment in the U.K., in a city of 300,000, the best I can get is 0.25/15Mbps - for £80 a month. It craps out completely every evening, down to 200kbps or so, as everyone streams netflix at potato quality.
I often wonder at observations like this whether it's the result of massive subsidies or massive over-subscription of the infrastructure. How is actual bandwidth/jitter on that line? Romania also has bold claims about infrastructure penetration, it's a fair example of somewhere I'd have good reason to doubt their credibility.
A stable loss/jitter-free FTTP connection at 50 GBP/mo. is very much value for money compared to an equivalent line featuring loss/latency/jitter at even a tenth of that price and with 10x the claimed line rate.
Separately, I have put off gigabit installation numerous times over 6 years simply because I can't really benefit from it on contemporary WiFi.
(Also yes, on re-reading my comment I realize I am a bit jealous of your setup in Portugal ;)
And for telecommunications, which are definitely in the "utility" category now, subsidies for the up-front capital costs are warranted IMHO, especially as it can act as an economic accelerator.
As for contemporary wifi, that's why I had my house wired for Cat 6 a few months back - I've got 3 Wifi 6 access points as well (PoE powered, ceiling mounted) and can saturate my 500/70 FTTP connection.
I think the key thing here is that probably yes, however it is essentially once off. Not that expensive to keep shooting photons down the fibre once its in the ground.
So from a country perspective that's a pretty grand deal compared to say farming subsidies that you need to do annually.
remote refid st t when poll reach delay offset jitter
==============================================================================
ua.pool.ntp.org .POOL. 16 p - 64 0 0.000 0.000 0.000
*62.149.0.30 .GPS. 1 u 353 1024 377 0.702 0.063 0.105
+91.236.251.13 105.240.56.33 2 u 40 1024 377 3.825 -0.068 0.466
-195.78.244.50 91.198.10.4 3 u 171 1024 377 17.445 3.617 0.251
-162.159.200.1 10.132.8.165 3 u 615 1024 377 0.460 2.218 0.115
-162.159.200.123 10.132.8.165 3 u 806 1024 377 0.489 2.212 0.104
-91.236.251.12 105.240.56.33 2 u 46 1024 377 3.879 -0.522 1.551
-91.236.251.129 118.188.39.164 2 u 58 1024 377 5.025 -0.056 0.293
-193.106.144.7 31.28.161.68 2 u 368 1024 377 6.991 1.203 0.120
+31.28.161.68 .GPS. 1 u 961 1024 377 0.543 0.152 0.134But this is on "home-network" technologies - they have their equipment distributed over buildings and have not central core, so this is cheaper, but lacks some things which seem normal for telecoms.
It is very reliable, easy achieve less than 5 hours of downtime per month. Most downtime because somebody steal equipment from attic, or somebody tried to steal fiber cables, thinking it is copper.
Example trace (skipped not necessary info):
traceroute -q 2 -w 1 --resolve-hostnames www.amazon.com
traceroute to d3ag4hukkh62yn.cloudfront.net (52.84.115.95), 64 hops max
2 94.244.0.210 (rusanovka-uaix.elan-ua.net) 0.519ms 0.432ms
3 193.239.72.201 (ae2-166-gw0.g50.kv.dataline.ua) 0.556ms 0.550ms
4 10.204.215.47 (10.204.215.47) 45.048ms 8.404ms
5 193.239.74.62 (amazon.dataline.ua) 27.452ms 27.259ms
6 * *
7 * *
8 150.222.27.44 (150.222.27.44) 41.083ms 41.283ms
9 150.222.27.197 (150.222.27.197) 40.665ms 40.495ms
Example ntpq -np
remote refid st t when poll reach delay offset jitter
==============================================================================
ua.pool.ntp.org .POOL. 16 p - 64 0 0.000 0.000 0.000
*62.149.0.30 .GPS. 1 u 353 1024 377 0.702 0.063 0.105
+91.236.251.13 105.240.56.33 2 u 40 1024 377 3.825 -0.068 0.466
-195.78.244.50 91.198.10.4 3 u 171 1024 377 17.445 3.617 0.251
-162.159.200.1 10.132.8.165 3 u 615 1024 377 0.460 2.218 0.115
-162.159.200.123 10.132.8.165 3 u 806 1024 377 0.489 2.212 0.104
-91.236.251.12 105.240.56.33 2 u 46 1024 377 3.879 -0.522 1.551
-91.236.251.129 118.188.39.164 2 u 58 1024 377 5.025 -0.056 0.293
-193.106.144.7 31.28.161.68 2 u 368 1024 377 6.991 1.203 0.120
+31.28.161.68 .GPS. 1 u 961 1024 377 0.543 0.152 0.134BTW could somebody give link, on what template engine uses HN, so I will post structured text. Thanks in advance!
very much. In early 2000 I consulted for small defense contractor trying to win bids providing internet to rural Spain. They made me cobble together an absolute garbage solution of slow early Sat (the ~1 second latency kind) internet with Wifi last mile. It barely worked for anything other than mail, browsing web was pure suffering, gaming was out of the question. Thankfully this venture went nowhere, but even with very expensive Satellite uplink finances still made some sense considering the remoteness of locations.
Must admit, your suggestion is very close to reality.
In reality, need complex - good basic education + accessible from early school computers + low level of income in country, so smart people must figure out, how to flee out of poor.
For example, in Brasilia, all people play football (soccer) at least once per week (many play every day), and they have best football team in world.
Another example, China was so poor, they have not seen home computers before late 2000s, so now they have not hackers culture at all, and so their startup system looks weird.
Another example, I grow in Ukraine, with access to ZX Spectrum, and now I'm here :)
Would take a 24 month contract though so practically I'm just going to sit tight on my 1gig
But yes, if an area is in a "not spot" you're stuffed - which is why things like B4RN took off.
Luckily there are a few new ISPs popping up now.
The house we stayed in is now in the starlink queue.
I've just stuck an external antenna up and we're getting a decent 4G signal from EE so that will have to do for the time being. It's peaking at 50mbs with a bit higher latency, but the upload is 25x faster than the landline which does make for far more responsive usage.
Annoyingly, there's a full Fibrus fibre rollout about 10 miles south, but apparently there's no money in running a line up the main trunk road between the two major population centres for this area.
I switched over to their FTTP offering recently and, while it’s not super fast compared to what others are claiming, having a router with a weird intermittent hardware issue reminded me of how good it is to be able to immediately jump on a call with technical people who actually care.
- I now see, 100mbit for one person is enough, I very rare see it is limiting factor (few times per month I doubt), but I can't imagine how I will utilize gigabit :)
* Usenet - easily maxes out 1g
* I've got a hypervisor here so lots of ISO templates, docker images, package pulls
* I host various things from home. Seeding (actual) linux images, sheep-it render server, crypto node
* Game file downloads - steam etc.
So definitely a luxury, but not entirely pointless
Ah, yes, I'm now limited by small drives (I have 256G SSD and 1000G HDD) :)
Before, when paid per mb, I created accounting system, and seen very clear, after my hdd seem full, bandwidth usage dropped.
But something broken in dependencies, and I moved from FreeBSD to Ubuntu, and on current unlimited channel it was not need to account, so I just don't bother, but I see difference when need to download ISO :)
Now I just for curiosity thinking about create accounting, make smart reports, etc, but on Ubuntu for this have to do much more work and much less art than on BSD.
It is still surprisingly thin...I set the limit at 5% of my pipe...and I don't think it has ever been hit. Not sure why but nobody is pulling down even close to what I could push
Sheep-it on the other hand...(think crowdsource blender renders) does actually push gear hard.
>But something broken in dependencies, and I moved from FreeBSD to Ubuntu
Do have a look at proxmox if you can.
Proxmox + ansible was a paradigm shift for me. (Proxmox has a CLI so you don't need any funky third party module anything...you can just deploy VMs/LXC via straight CLI)
Sure, I use virtualization when possible, but I prefer Docker.
But for me now problem, that some vendors prohibit usage of their devices in virtual environment, so I have to stay semi-desktop. For example, nvidia official drivers for desktop (game) cards, don't work in virtual env, and server cards are slow in games and have worse performance/$ rate.
In some distant future, I believe, will buy dedicated server machine, but now using desktop, to save money and space.