That's insane, how are they covering their infrastructure overhead, offering 10Gbit connections for 40$ a month lmao? That is normally enterprise level for businesses with +300 employees.
If this is true I might have to consider a summer home in Sweden. I have never heard of anything like this, literally shocked.
It wasn't 100% correct, the reduced price of $40/mo is only valid for the 1st year, 2nd year is charged $50/mo. What a rip off.... ,)
10kr is slightly less than $1.
Once the fibers are in the ground over the whole country it is mostly only about changing the hardware in the endpoints once the hardware actually exists.
Don't bother. Let a router route. Get dedicated APs for wifi. I use ubqiuti APs for my wifi. I use a Friendly-Elec R4S for my router with OpenWrt on it.
The age of All in One routers is better served by dedicated equipment. Especially if your connection comes into the house in an awkward place. Having backhaul wired APs to serve centrally is far better for coverage.
Heres some articles that you might want to read and have a think on. The Ars one is older (2017) but alot of what he says is still relevent today and points out a fair bit that you will also have to consider.
This one is from 2020 and explains about wifi ranges and limitations.
Same in Bulgaria. 10 Gbit around 51 USD monthly for home users, but the question is do you really need it? I recently changed ISP. I was on 100 megabits and was wondering a lot if I should go to 500. Well the difference was 3 USD, so I will try it for the next 2 years, but do I really need it?
Yes and no, for quite a number of apartments a single AP is just enough, and the wrapping everything, the router, the modem the AP and the VoIP basestation into a single unit can be quite attractive to some, it certainly is more efficient energy-wise to run a single slightly larger power supply to supply all functionality on one device compared to haven a bunch of independent devices each with their own power supply unit.
Don't get me wrong I have a lot of sympathy for splitting functionality into independent devices (I operate an independent VoIP base station so I can choose my router without having to bother about VoIP capabilities), but I also accept that there is a large demand for smallish power efficient-all-in-one routers.
Personally I operate a turris omnia I got during the indigogo campaign years ago. Overpriced? Arguably yes, but also a nice package and fast enough for my current needs and desires. Since I got a "power meter" only recently I can not yet tell you about power consumption but measuring that is on my todo list. 
Not everyone needs VPN, SQM and a gigabit ISP connection.
For a user with 100-300 gigabit connection who needs good Wi-Fi can get a router for $50-$100 that covers all needs. For those who have a bigger house with brick walls needs some APs in addition.
That's the most common use case.
For example. I have a 100 megabit connection, because I don't need more. My neighbor with a big family has a 300 Mb connection, he uses a router with 2 additional APs. We don't need anything more.
I understand that argument, but in my opinion that is not the Linux spirit. Getting a commercial solution that could easily be outperformed by a far more versatile and cost effective DIY build is not how we do things.
That is the essence of Open Source. Defying corporate moneymakers by literally producing our own solutions that almost exclusively turn out better anyway.
But on the other hand, I can also see why some people might not have the time or energy for such a project. Buying something quickly and it works has its merits too, don't get me wrong.
I just feel like we are all part of the Open Source community, which is the most powerful tech force on this planet. Projects like OpenWrt almost demand to be honored with custom builds lol.
ARM64 based router + separate access points is what I'd recommend as well. If you go WireGuard, you won't need the extra power x86_64 brings to the table to hit higher speeds with a VPN.
But openwrt runs on lots of different devices that suit differing needs.
All in ones are great for people with limited space, limited needs and wanting lower power consumption. They can be cheap or expensive and vary massive in performance.
Openwrt is, in my opinion, is about freeing devices from the restrictions the oem put on it and making it more useful.
I do agree that the crazy expensive all in ones are of questionable value, but some people need what they offer, like 10gbe ports or fibre sfc etc
Personally I run a x64 thin client router (pulling 10-15w), with separate access point and switch - but a more complex solution like this isnāt for everyone.
With x86_64, it's all about proper hardware selection. Yes, your old gaming rig will draw 70-120 watts idle, but well selected x86_64 components can be very power saving.
Case in point (all values measured at the power plug):
While the trend has slightly changed again, with Intel trying to counter AMD's ryzen CPUs and concentrating more on performance (again), rather than good idle power consumption or heat/ noise, the same can be achieved with more modern x86_64 hardware - but you have to be picky, mainboard and CPU-graphics (no dedicated graphics card) being the main areas of attention, in general less is more (less features, less slots, etc.). Ironically my ARMv8 based AP needs more than twice the power than my router (four radios take their toll).
TDP is a useless metric for this task, your router will not be at full speed 24/7, even a busy workstation will be spend more time in its idle modes than running at full power, let alone a home router. But for the examples above, both j1900 systems will rarely go beyond 20 watts (shortly), the c1037u won't cross 30 watts - and yes, the haswell based could reach the 3-figure range, if you push it with non-routing tasks (it won't bat an eye or clock up for anything you could throw at it over the internet).
Disclaimer: the values above are established on my rather dated hardware, which does the job nicely - I don't have comparisons with more recent hardware.
For me, the UPS tells the power. But there is some more things connected on my measured 20W so I donāt know specifically how much only the core network hardware burn but it is less than 20W.
Well, once you rolled out a fiber network, especially a passive optical network (PON) which you might finance over decades anyways, all that differs between 1 or 10 Gbps is really the optics at both ends. On the ISP-side the OLTs nowadays offer multiple standards concurrently over the same fiber tree (this works because 10/10 Gbps XGS-PON can use different frequency bands than "legacy" 2.4/1.2 Gbps GPON and so the ISP only needs to supply those customers with 10 Gbps capable CPE that actually require it) while CPE typically are still supporting mostly one standard (but there are CPEs available with flexible optics using SFP modules, like AVM's fritzbox 5530/5590 that come with both a GPON module as well as an AON ethernet over fiber module so it should work on all FTTH links in Germany).
Point is, 10Gbps capable hardware is sufficiently cheap, that ISPs can afford to produce high speed plans for low costs if the local market demands low costs. You can find examples of this all over Europe, like in this thread mentioned Bulgaria and Sweden, but also France and Switzerland, and likely more.
The situation in Germany is less advanced, the bulk of current FTTH roll-out uses GPON (and a sizable but shrinking fraction 1 Gbps AON), but even here the first niche ISPs start to offer XGS-PON based plans >> 1 Gbps and the incumbent is at least experimenting with this in its laboratories.
However there is a challenge for German ISPs in regards to "10Gbps"-XGS-PON, in Germany contracted and advertised rates are defined as rates users can achieve as net goodput on IP/TCP/HTTPS speedtests, and XGS-PON with FEC only offer something like 8.6 Gbps gross rate, which only allow around 8.2 Gbps of theoretical goodput; so according to rules & regulations ISPs can not advertise with the "magic" 10 Gbps number, so I expect that ISPs will introduce something like 2 and 5 Gbps plans instead) But I digress
I can beat that: My Sophos XG 86w draws 7.5 watt at regular operation, and that includes the wifi card it contains. By all accounts it is complete overkill for a router, it PPPoE's and NAT's even an FTTH gigabit line without looking up from its crossword puzzle.
Even though it is only 3 years old hardware, it is now EOS and consequently only available second-hand, and as it's nearing EOL, IT departments are giving it away for cheap because the value to them really lies in their OEM software, not in the hardware. After a little bit of patience, mine cost me ā¬75, but they (or similar machines) can easily be had around the ā¬100 to ā¬120 mark. So I don't really get where the "$500 to $600" price tag complaint comes from.
too good to be true should make you worry - you know this saying right. don't get fooled by ISP marketing: 10G is the physical connection layer most probably. unless there is solid evidence that:
a) they always provide you this bandwidth for actual usage, i.e. anytime, anywhere, regardless of other users (who might share the same network segment, so no over provisioning),
b) their upstream can keep up with 10G ( times concurrent users), otherwise this only means you can connect to your peers within the ISP network via 10G,
it is just a marketing bs.
not to mention the your own network should be also 10G ready ... still not too many affordable 10G switches for example. and other ends you connect to should also support higher potential bandwidth (all the way to your router through the whole public internet ...) to achieve more download speed.
so imho 10g sounds nice but probably not real ... the physical link is 10G but the whole service likely not. but test it and prove me wrong, and you'll be a happy customer for only 40-50$.
The chance per connection to get a download speed of much more than 10-15Mbps from any upstream server is in reality pretty slim to non existent.
A speed test is fine numbers but useless data.
That is an unrealistic demand. The whole internet is a shared medium, so "anytime, anywhere, regardless of other users" is just not a useful concept here.
Exactly my point, their upstream is shared, as are the internet connection from most servers. All in all it is probably safe to say the the aggregate contracted rate of all edge devices exceeds the aggregate connection rates of all servers by several orders of magnitude, so over-subscription is a fact of internet life and not an exception.
However, i different area there are different rules about what rates ISPs can advertise and offer and what remedies exist if those rates are not actually delivered by the ISP. IIRC ISPs in the UK can only advertise with average rates their users reach on actual speedtests. And in Germany there exist an official speedtest plus detailed regulations how ISPs need to inform their users about the speed they can expect, how to confirm that, and what remedies exist if the ISPs does not deliver the contracted rates according to the interpretation of the national regulatory agency; in short users can either demand an immediate cancellation of the contract (interesting if there are alternative ISPs) or they can adjust their monthly fees proportionally to the degree the ISPs fulfills the contracted rates.
However none of these methods (UK or Germany) boil down close to "anytime, anywhere".
All true, however ISPs deploying 10 Gbps plans are also deploying CPE that nominally are capable of those speeds. And for many users it probably does not matter whether they achieve the top end of the speed as long as things are considerably faster than with the previous plan. Also increasing one's own home network to 2.5/5/10 Gbps is something end users are responsible for themselves, not something to "blame" ISPs for.
Mmmh, I only have a "lowly" 100/40 Mbps link and see routinely download speeds in excess of 15 Mbps. I am not saying anything close to 10 Gbps will be achievable (let alone as single flow transfer over the internet) but 10-15 Mbps seems too low a threshold from my subjective experience.
Yes and now, if you mostly use your link for bulk transfers speedtests to selected severs can be quite predictive of what aggregate throughput you could expect; I agree however that they tell very little about suitability for interactive usage.
However, Ookla just added delay numbers for the upload and download measurement phases which give at least some feel about responsiveness of a link under load:
See the three numbers reported for Ping ms (idle, download, upload)
(test performed over a non-home 1/1 Gbps ethernet link shared with other users, so ignore the reported rates, this is just an example to show off the new latency under load numbers).
You just answered your own question.
My thoughts exactly. Personally I simply don't have the time anymore to assemble custom hardware builds. No time for researching and selecting components, no time for finding cheap suppliers, receiving packages, screwing stuff together etc.
At some point the spent time is far more worthwhile than the higher initial cost for an integrated solution.
that was my intention as well to point out that physical link and real service speed are two very different things, regardless how marketing is trying to mix the two (*). so, yes i know internet is shared medium with over provisioning as the common practice.
well, as i see it is quite easy to be compliant with the magic word: upto all contracts i've seen defines the minimum, not the average nor typical, and definitely not guaranteeing the advertised maximum (a.k.a upto).
it was not blaming ISP just pointing out having 10G/1G ISP service with 100M home network will make no sense. but to match with 1G ISP service upgrade home network to 1G is easy and affordable, but matching 10G service is not affordable imho due to still expensive >1G switches.
nevertheless, if you can afford 10G and it will be >1G in reality, you should do it. in short my point was to challenge marketing.
