I think things like this are great because they will finally start to create pressure to commodify >1gbps networking in the home. The industry has some work to do here.
I've been on Sonic 10G for almost a year now and LOVE it, but it's definitely been a sore spot to get things set up to expect those kinds of speeds - prosumer 10G switches are vastly more expensive per-port, wired consumer devices don't typically support 10GbaseT/SFP+, 2.5G/5G switches aren't broadly available and commodified (e.g. UniFi has limited offerings here) and WiFi6E is still mid-rollout (almost no client devices currently in market yet) meaning that clients can't reasonably expect >1gbps of goodput, even with a good link from a modern device to a modern AP. Then there's flakiness: when things get hot in my garage, my 10G switch just stops working. My Thunderbolt-to-10GbaseT adapter for my MacBook runs very hot. Lots of sharp edges here.
The more consumers are buying 10G equipment for their 10G home links, the faster prices will come down and reliability will come up - not just for 10G but also for 2.5G/5G equipment. Hats off to EPB for paving the way for 25G and keeping vendors diligent in mapping out the next generation of their equipment.
10GBaseT is the wrong solution, period. As you've found out it runs hot (somehow I doubt all the fanless consumer trinkets are going to last very long). Not that Ubiquiti is a good example – even with their 1G kit they had issues with thermal management. Both the ER-X and ER-L ran hot enough to cook themselves (especially the early ones with the tiny SD cards). True to form my ER-Xs get flakey when it gets slightly warm. Fiber is the way forward and hopefully getting multi-gig connections to the home will start to bring the pricing on some of that kit down to reasonable levels.
FWIW I've also got Sonic's 10G offering and yeah it works well enough. Something is causing intermittent problems that I've not had with their ADSL or 1G fiber offerings – I'm starting to suspect the 1G TP-Link switch they tossed in. For now I just live with the periodic interruptions.
Edit: Certainly I'm not planning to upgrade any of my gear to 10G as I simply don't need the bandwidth and Sonic's offerings are really geared towards web browsing more than anything.
In my extremely limited home 10G experience so far, SFP+ DAC modules have seemed relatively inexpensive with low(er) thermals than 10GbaseT (subjective SWAGs here, not detailed thermal measurements!) and are a great, cheap way to connect a desktop to a 10G switch. Kinda makes me fantasize about a laptop with an SFP+ slot, TBH...
I was admittedly just too chicken to wire my home with fiber (nervous about kinking it) so used plenum CAT-6A for my mid-length runs (~50ft). Perhaps there are classes of "tough fiber" for this use case?
I gotta say, I LOVE Sonic. $40/mo and I actually get 6gbps+ up, 6gbps+ down real measured throughput, and unmetered/unfiltered. Great support and a great CEO, Dane. Just wish I could pay a bit more for a static IP, like I did with AT&T Fiber.
So the last time I looked I only found one person complaining that their ER-X runs hot to the touch (plenty of "it seems kind of flakey" comments tho). Mine become unstable once the ambient temp gets much past 75 °F. The ER-L is well known to run hot to the touch at idle.
When I opened up one of my ER-Xs to get a the serial header I noticed that it had a much sloppier application of thermal compound than the one pictured in the instructions I looked at. I'd totally believe some units will be more reliable than others.
No static IP on fiber connections (like they promised initially), no native IPv6 (although maybe there is on the DSL products), outbound SMTP blocked. For a while there was a tool you could use to adjust your DSL profile to skew it towards upstream bandwidth, but that's since been dropped. No more m-bone access. Stuff like that.
Sonic's shifted from offering a full featured internet connection to a slightly neutered dumb web browsing pipe.
There's more to Internet than downloading data straight to disk.
Families can have multiple kids taking online classes while both parents are also working remotely. Many families have multiple devices per person that can be streaming, downloading, OR UPLOADING (commonly the aspect shortchanged in residential Internet) which, at times, can easily saturate a 1Gb connection.
We always need to advance because others are too, and services demand greater speeds in returb. It's a bit of a race, but no need to claim it's useless. We can do things now that were unthinkable before due to such advances.
>Families can have multiple kids taking online classes while both parents are also working remotely. Many families have multiple devices per person that can be streaming, downloading, OR UPLOADING (commonly the aspect shortchanged in residential Internet) which, at times, can easily saturate a 1Gb connection.
My 37mbps can do that for 2 people constantly working from home and most of the time on meetings.
My guess is that most people will just use whatever router is supplied by the ISP, and those are going the be the cheapest China has to offer. Even if it is Wifi6, that will max out at around 1Gbps. Some might run Ethernet to their office or consoles, but I'd guess that most will just use wifi. There's also the issue of the networking gear. Home users will just buy 1Gbps equipment, or slower, if that's what fits their price point.
For future use, and businesses it is nice to see speeds above 1Gbps being rolled out. For now though, promising 25Gbps to residential is easy marketing. The ISPs knows that it will barely be used.
20 kids x 25mbps streams each is still only half a gig. You are way off here. The only way you saturate a gig is if you download big games from Steam, etc. If each of your kid is doing that you would probably run into fair use limits and have to set up a central PC for catching downloads and stuff but that's still a niche use case.
> which, at times, can easily saturate a 1Gb connection.
The only thing that can saturate 1 Gb in practice in my experience are large downloads.
Video conferencing, streaming etc. uses well below 100 Mbps (even 4k Netflix seems to be 25 Mbit/s). You'd need dozens of such applications running at the same time to come even close to saturating Gbit.
This may change over time, but right now, there is very little need for going beyond 1 Gbps for households. And honestly, I think the change will take a long while, because 4k is still not ubiquitous, anything higher than that isn't much of a thing, and even the next step after 4k (assuming it needs 4x the bandwidth) will still be easily handled by a Gbit line even for large households.
A 100 GB game download taking 14 minutes with a saturated line is not going to be a reason for most people to pay for faster hardware all along the path (router, cabling, NICs, ...).
Replying to my own comment. My examples weren't the greatest as I was traveling while composing it.. I agree that my examples would not push a symmetric 1Gb connection to its limits.
I was trying more to make the point that folks were giving arbitrary examples like SSD speeds when there are other kinds of usage that have other limits or could potentially change with time.
If residential speeds only target the "average" case, that'll artificially slow advancement. Many places outside the US have much faster speeds than are available here and we should not invent reasons to slow the shift to faster access. At least it shifts the needle a little and puts additional pressure on the lazy ISPs that still offer 90s-era ADSL speeds.
My family of 6 would stream Netflix, game and I'd be pretty heavy handed myself all on my router tethered to my 4G connection while we were switching providers. We didn't notice anything surprisingly. Three was the provider at the time.
You may be off a bit. A gigabit is only around 125 megabytes a second. A low end consumer SSD is closer to a 550 megabytes second, or ~4 gigabits. A high-end NVMe SSD will get you 3 or 4 gigabytes a second. 30 to 40x a gigabit!
Grandma running pirated Word 2007 on an ancient dust-clogged Dell laptop that her grandson upgraded with a cheap 256GB Lite-On mSATA SSD stolen from a throwaway work computer is probably only getting 125MByte/sec and certainly lower than that due to the plethora of IE addons constantly writing to it.
I'm now imagining this ancient grandma writing 125 MB/s into her pirated text editor, just cranking out volume after volume of cheap detective and romance novels. Later, after her death, it is discovered that this lady was the ghostwriter behind all of the top 50 bestsellers of the last 30 years. She would write faster too, if it weren't for that crappy SSD her grandchildren set her up with.
Under ideal circumstances, yes. When say, torrenting, or anything else with non-ideal file access patterns real world speeds are not going to be nearly that high, especially on a consumer-grade drive.
Stress tests on consumer SSD's consistently show 300-500megabytes/s.
The whole point of an SSD is that it has ridiculously high available iops, so what you are saying literally does not make sense.
a consumer SSD will have about 50k iops at that 300-500megabyte/s . The access pattern will not practically matter at all until you saturate that or get close. The drive is effectively constant time until you hit the limits of the interface, or at least get somewhere near it. A consumer will never get near it.
So they can expect, and consistent testing shows, they will get 300-500 megabytes/s in real world conditions, and even harsh conditions.
I actually have no idea why you don't just say "yeah, i was wrong" and move on.
It's clear you are wrong about this - there is no data to support what you are saying.
It's also totally and completely orthogonal to a more rasonable argument - they don't care about the speed anyway.
IE arguing they don't have equipment that can saturate 1gbps is silly - they clearly do.
arguing they wouldn't care either way because the speed difference doesn't matter to them is more reasonable.
Yeah, I completely agree with you there! The average person doesn't need gigabit speeds. The only reason I even upgraded from 100 megabits to 300 is because it was literally 5 bucks more a month.
This was my experience when I had a gigabit connection. I could download games from steam as fast as my ssd would write it.
Which was pretty awesome, but I'm not often pushing that sort of bandwidth.
The counter argument being that without it being a commodity, we don't know what might be invented if it were. We could miss out on, or delay lots of really amazing innovations.
That’s not my experience at all. I had 10Gbps internet in Tokyo. In normal browsing there wasn’t much difference between using wifi which I normally did and connecting by 10Gbps Ethernet. But downloading on steam was one case where the difference was amazing, I could download a AAA game in 1-2 minutes instead of 10-20.
I find that Steam decompression pegs one core of my CPU and downloads get stuck at 550-600 mbps. It would take a lot less time if it didn't compress anything.
Even as a regular consumer I see the benefit. I buy a new game, and its typically something like 100GB. Gigabit internet is the difference between waiting 15 minutes and 1 hour to play my new game. This is not an uncommon use case and the difference is quite large.
Right - the popular sentiment on my smallish town Facebook group is people dropping their 300mbps cable since Spectrum is an awful company to deal with in favor of cell-based (T-mobile or Verizon) home service. 90%+ of people in most areas don't need internet any faster than Netflix requires, and nobody is running ethernet so WiFi speeds are going to be the limiting factor in internet speed for most everyone else.
I went from ~250Mbit to Gigabit recently. I don't notice it on my wireless clients at all, but it is really nice on my main PC, which is tied directly into the router via a good 'ole cable. I've seen actual downloads speeds flirting with 100MB/sec, which is really nice when downloading some of these massive 50GB+ games.
I'm a deep computer nerd, I play lots of games, with a family watching streaming on 2 to 4 devices regularly, I buy a decent amount of games, download terabytes per year...
I'm just fine with my 300 mbps connection. Even when it drops to 20 mbps occasionally it's not the worst experience.
I'm not sure what it is that is different about me than my peers.
I'm right there with you, as long as that 300 mbps connection is symmetrical, and the ISP is carrier neutral. I didn't know how big a deal neutrality was until I switched from Suddenlink to a small company called Vexus. Gigabytes of nightly backups went from hours to 5 minutes, and downloads from certain companies improved drastically.
Your SSD comment aside: not everyone lives alone; during the lockdown we had four people using the network simultaneously during the day, plus of course some of my other machines using the network. We have friends with three or four kids, which would have magnified that demand.
Gigabit can support ~25 simulatenous Blu-Ray streams (and that's at the full 40Mbit).
It wasn't all that long ago (~10 years) when the company I worked for had only 200Mbit to the internet, and that supported ~300 people pretty comfortably.
Since every consumer fiber plan is best effort, I'll get better speed by using faster plan even though rated speed isn't needed. I get about 700-200Mbps on my gigabit fiber for download. I'm fine with 1Gbps but 200Mbps is sometimes slower. Hopefully 2Gbps is constantly available on 10gigabit fiber.
An SSD nowdays has something like 3500 megabytes/second, a 1gbit connection can (roughly) provide at most 100 megabytes/second, so we are not even close to use SSD write speed at full capacity.
That being said, downloading from Steam is incredibly different with a gigabit bandwidth.
10Gbps wont ever be consumer grade. There is nothing on the roadmap which suggest we could bring down the cost of 10Gbps Ethernet to say within 2x of 1Gbps Ethernet. 2.5/5Gbps have a much better chance, and it will require some market narrative to push for that. Such as WiFi 7 ( 802.11be ) getting 10Gbps wireless speed but stuck with only 1Gbps Port. It doesn't matter whether that is true in real world, but with enough Social Media push and marketing, we could push vendor to incorporate 2.5 and 5Gbps. Then we may have some chances of seeing 2.5/5Gbps replacing 1Gbps Ethernet.
> prosumer 10G switches are vastly more expensive per-port
They are, but they aren't out of reach - i recently bought 2 Zyxel managed switches with 3x 2.5/5/10G and one SFP+ ports alongside 8x 1G ports for 200€ each. That's not cheap, but it's not super expensive either, and most consumers can get away with unmanaged switches. The unmanaged version has 2x2.5 and 2xSFP+ and comes for 120€.
The kicker I've found when looking at proper enterprise networking gear (NOT anything from the Ubiquiti trash lineup) is most of the cost on top of an already high hardware cost is in software licensing for the router / switch itself. Did you different routers etc?
Fortunately, fiber cable runs are really cheap. Fortunately with sources like FS.com optics and transceivers also aren't too expensive.
I think that wired ISP bandwidth missed out on marketing/branding (which may also be intentional). The thing is that the average consumer is woefully misinformed. You don't even need to look that far. A comment downstream mistook gigabit with gigabyte. With cellular networks, everyone and their dog is clamoring for 5G. Wifi too now has numbers like wifi 6 and wifi 7. So my conclusion is that we need numbers but not units. Units confuse people.
The problem with municipal Internet is the town will build it out once, and then never upgrade the technology while it slowly becomes obsolete </s>.
Seriously though, municipal fiber is fantastic. I've been an observer to friends having conversations about Internet provider woes, while I just sit there shaking my head. I've only got 1Gb but I rarely care about the bandwidth, consistent <15ms ping to close data centers, there's no data quotas, no yearly fuckery where they dick you around with the price and you have to wait on hold to threaten to cancel and commit to a longer term, and I haven't noticed any downtime in years. It just works.
The OP story is literally the municipal provider delivering 25x faster service over 12 years. The "never upgrade the technology" provider is usually going to be monopoly Comcast, because they know your other option is a can and string.
So true. Almost 10 years after DOCSIS 3.1 was released, we still don't have better upstream speeds. I wonder, is there a cable company anywhere that actually does support DOCSIS 3.1 upstream?
I don't know what a book could do for you, as it would be too literal/linear regardless of the meta. What I do know is that understanding sarcasm requires critically evaluating everything you take in, and being keenly aware of the context around it. (Hence it being good to tag impersonal written communication - if we were meatspace friends you'd already have a strong assumption that I wasn't a Ma Bell lobbyist drone). And hearing corporate marketing speak parodied sardonically makes you less likely to just accept it when you're earnestly inundated with it. I'd also say this topic is closely related to "security thinking" where one critically examines the converse/implication of every statement.
I live in another TN city rolling out municipal fiber. I live in a middle class area of town. I've been asking for 1G for years, but ATT keeps telling me "there's not enough need in that area of town". That changes as soon you move to the higher-end neighborhoods though. I guess people who live in the expensive zip codes are the only ones who need 1G. EPB will offer internet to ALL residents, not just the rich of the cities.
Since the city has started the fiber rollout, all of a sudden it's now feasible for ATT to offer fiber. Go municipal internet!!!!!!
Wow this reads like an announcement from a different timeline (like the one that was promised when fiber was promoted in the early 2000s). 25Gb/s symmetric? In the middle of Tennessee?
Meanwhile I have traffic-shaped Comcast and no fiber in sight in Silicon Valley.
My parents in (poor) rural Michigan will soon have county wide fiber internet provided by the local electrical co-op. I'm sure the support will be semi-local and so will the technicians. So you'll actually get people that at least pretend to give a shit about your problems all for a better price than whats available in the city from Comcast.
And yes, they and ATT lobbied extensively to block communities from building out their own ISPs because they didn't want to compete. Michigan has a law in place where a municipality can't build out a service if at least 3 companies are willing to bid on running 'high speed' internet service in a town.
I have WK&T and have symmetrical gigabit w/ 24/7 chat support in a town with a population of 300 in TN. Spectacular service, but I did sit on a waiting list and pay $1000 for installation. My bill has gone down over time (WK&T is a co-op.)
I have only ever had issues for a period of 2 months during peak times when my traffic was being routed through another ISP that I would always get ~0.2% packet loss from. WK&T helped me identify some routes that were unaffected to set up a tunnel to work around it for the few times I was affected.
its weird, I live in a tiny little town with about 400 households and I now have access to 1G symmetrical fiber for about $80/month - but we (the town) had to pay for the fiber build out with increases taxes and user fees...well worth it in my case. I would have been OK with only Comcast as previously it was just 3Mb Verizon DSL for almost my entire WFH career (20+ years).
A cynical person might say that large telecoms company aren’t interested in monopolizing your market and so you are free. I’m astounded by how bad/expensive my internet is in massive cities that should be easy to connect. I live in an apartment in LA and have very few options. I lived in a loft in downtown before (5 years ago) and the building only allowed DSL. I have a business connection in Portland and pay around 100/month for 15Mb.
I'm in Brooklyn by prospect park and only have 1 ISP choice for my apartment. And it's coax based, so every plan caps at 20Mbs upload. It's that or satellite internet. I miss having fiber.
Yeah, seems some rural communities are getting fiber faster. My house in northern GA far from Atlanta has symmetrical(?) fiber. It gets 1000 up and 1000 down.
A big part of the reason for this is that rural areas/exurbs get heavily, heavily subsidized.
A few years ago my local ISP (Cincinnati bell) literally stopped their profitable roll out of fiber in the city for the better part of a year. They got a huge subsidy to put fiber in the exurbs around the city, and shifted all of their people to those areas. Go figure.
Chattanooga is actually almost in Georgia, and not too far from Atlanta.
But yeah, it just goes to show that with the right opportunity and motivation, you can really do some cool things in places you wouldn't expect. I used to work remotely for a company in Chattanooga, and I was always blown away by the internet speeds whenever I visited the HQ. I remember downloading a 300mb file in 10s, in 2015. The story behind how it all came to happen is quite cool too.
This has been going on for a while, comments like this and the recent Tennessean article/debacle are a bit eye opening to the naivety of technology and accomplishments outside of major hubs.
“real estate is expensive and NIMBYs don’t want it”… the end.
(Also the story of why Hyperloop is pointless. technology isn’t the problem that needs solving! Fast rapid transit tech has existed since the 1980’s and it’s not the cost issue.)
That, and don‘t let the fox in the henhouse (namely, if you decide that contracting to consultants is a good idea, you do need in-house staff to make sure they‘re just not perpetuating an endless gravy train, and other consultants are not a substitute for this)
It is cool, but I imagine it's expensive. Tried their site with a random address, and 10Gig is $300/month...their 1Gig is $68/month. 25Gb isn't yet listed.
Haha, I work remotely now in a rural unincorporated area near Knoxville, and we get AT&T fiber at 5 Gbps, with the local municipality rolling out 10 Gbps soon. When I was living in Redwood City, nothing like that was available at our address. I was astonished at the poor internet provider options in Silicon Valley.
It's a five minute bike ride on flat land from Google HQ (mostly along utility right of way) to housing with < 6mbit DSL.
Drive 10-20 miles, and you'll be in areas where AT&T decided to sell the lines to a bankrupt telco. Looking at the lines in those areas is entertaining. The telephone poles were installed by some ancient secret society named "GTE", and now have 20 degree bows. When lines loosen up and block traffic, the usually just tie them up on to some nearby tree branch.
If you look really carefully, you'll occasionally see fiber points of presence dangling precariously from this mess of caution tape and guy-wire.
It's not all bad news: I know of communities outside of telco right of way that managed to tap into one of those.
Last time I heard they were debating between 1 gig symmetric to each home or paying a couple hundred bucks (one time) per house to get something comparable to what you'd expect in Tennessee.
It's definitely a problem with incumbent monopolies.
Drive 10-20 miles, and you'll be in areas where AT&T decided to sell the lines
to a bankrupt telco. Looking at the lines in those areas is entertaining. The
telephone poles were installed by some ancient secret society named "GTE", and
now have 20 degree bows. When lines loosen up and block traffic, the usually just
tie them up on to some nearby tree branch.
The way I remember it GTE did not go bankrupt and instead became part of Verizon. Most of the Bay Area was PacBell (landline) territory, but there were a few pockets here and there where GTE had a monopoly. For wireless it was a different story with AT&T / CellularOne getting Side A and GTE / Verizon getting Side B with PacBell being relegated to the PCS band.
In fact I just took a quick peek at their Wikipedia page. GTE went bankrupt in 1933 and recovered. They were not part of the Bell system until 2000 when they were acquired by Bell Atlantic as part of the creation of Verizon.
Monopolies suck but GTE (and Verizon) were almost always better about building out higher speed DSL and fiber than PacBell/SBC/AT&T ever were.
I recently upgraded to 1.5gbps and realized a few things.
The downsides: the Cat5e cables inside my condo walls can't handle more than 1.0gbps; my desktop's network port also maxes out at 1.0gbps; the wifi router in my home can't do more than 600mbps even if standing right next to it.
But also, nothing I do needs this much bandwidth. I can download any steam game in the blink of an eye. I can watch streaming video at resolutions higher than my eyes can distinguish. Outside of a few high bandwidth edge cases, there's very little I need more than 100mbps for.
If you offered me 25gbps internet right now I wouldn't be able to tell the difference between it and what I have!
Someone tell me: what will I need that much bandwidth for?
What will you need that much bandwidth for... in the future.
Today? You don't need 25gig. Even if you're in a house with all the latest technology and a half dozen users, you probably won't ever saturate that line unless everyone REALLY tried.
But tomorrow is a different story. Ethernet above 1gig is becoming more common in the consumer world. Not super popular, but its starting to take shape. Wifi 6E has a theoretical maximum of just shy of 10gig. Wifi 7 (not ratified) is looking to be around 40gig.
With work from home and, in general, the world going more digital - there is a lot more pushing around of files. Do you have an iPhone or Android? Everytime your phone is on the wifi its backing up your photos/videos (and other stuff). That alone can slow down most peoples internet connections. Does your computer do backups?
Also once your internet starts to hit about the 1gig mark, you can treat remote servers like we treat LAN today. Want to have a server but don't want to have it at home? That's ok when you've got 1+ gig internet. Of course, it'd be even nicer if that was even faster... in theory.
Doing multiple of those things at once. You're watching a movie while one kid is downloading a game and another is watching something else? I'm saying it lets your whole family never have to interact with one another.
All your network actions are processing 5-10x faster. A lot of the little things you do are instant at 1GB that added up at 100MB. Medium files are small, and small file transfers are barely noticeable. It changes the feeling of any UI that uses the network.
At 10GB, once you have hardware that supports it, the same thing is going to happen again. Your medium-sized files will feel instant.
On top of all that, you save time on large downloads, huge downloads become not inconvenient, and your whole household can pretty much do all that simultaneously.
Right now, 25gig would only be seriously considered for a business, though.
I think the peer reply about treating WAN like LAN is the best lens.
Imagine all of your family/friends were on a local wired connection together. What would you do with that? Usually I don't find myself wanting faster bandwidth to interact with big services (Netflix/etc), but with other people or my own remote resources.
I throw around a lot of large VM images, which is definitely outside the average person's uses, but stuff like sending backups to your friend with the huge NAS, Steam/Parsec 4K streaming from that other friend's GeForce 3090 machine, etc...
Steam games is the biggest benefit I saw. Modern games are in the tens to hundreds of GB now so gigabit internet is the difference between waiting 15 minutes and 1 hour.
When your microwave, fridge, oven, tv, toaster, dish washer, dryer, tea kettle, and stand mixer are all playing you 4K advertisements for a new TV show or a new pair of Adidas, you will be glad you have 12gbps left to browse the web (which will by then require another 2gbps for all of the javascript, images, and 4K video ads needed to render a blog)
I noticed the same years ago. I have ~250mbit FiOS. I could upgrade to 1 gigabit for another $20 per month, but I don't feel that my life would be $20 per month better. Websites and Youtube wouldn't download any faster. Due to latency, work won't get any faster. I buy and install new games maybe once per year, but is saving an hour once a year really worth the extra $240 for that year? I don't think so.
I'm the only one at my house. My brother might come over, but that's just one low-end phone on wifi, which feels like a rounding error on ~250mbit.
When we bought a house, the best one ended up being literally a couple miles outside of EPB's service area and I had to switch to Comcast. I see their billboard every time I drive home, just to rub salt in the wound.
You have different parameters than I, because "the best one" has 25Gbps to the home, and not Comcast for an internet provider. :-)
I know what you're saying, though. I suppose it's possible to have higher priorities than bandwidth. Personally, I'd still have to give serious thought to how much I want $FEATURE if it means having to do business with Comcast (or Comcast aside, give up 25Gbps). Backyard for the kids? They can go play in the park. :-)
I love how so many companies offer work-from-home now as you can literally work from anywhere in the world as long as it's only in a major city with a high bandwidth connection.
Check out this map, which doesn't seem to include starlink. Click "min price" (to make the 25+ MBPS filter appear), then click "25+ MBPS". You'll see that rural houses are likely to have access to better Internet than ones in Silicon Valley.
To really light the map up, click "fixed wireless". I've had mixed luck with such ISPs, but they generally offer affordable plans that have better uplink bandwidth and ping latency than comcast.
Cool, but try to get 25Gbps on your mac. It's tough. We have our own 10 gig symmetrical fiber backbone connection and wanted 25Gbps on a mac in our film production facility. Thunderbolt maxes out at 40Gbps so you're already close to the max data transfer rate for a peripheral. We had to use ATTO's hardware which is bulky, a pain to configure on a macbook (you need to bond two interfaces) and honestly most of the time it's just easier to stay on Wifi and deal with 400Mbps throughput.
Not sure about windows, but USB-C maxes out at 10 Gbps IIRC, so I'd love to hear what folks in that realm are doing.
Please, Br’er Fiber, don’t throw me in the 25 Gbps patch. Anything but that.
Seriously, I would be happy to take up the challenge if someone is willing to provide such service.
More seriously, that bandwidth probably doesn’t go to a single computer — those of us with multiple people in the household working or playing from home can do more & better things simultaneously.
Yeah frankly I just hardline for 1Gbps and anything over like…I don’t know, 1TB? I just ship a drive. Anything beyond 1 or 2 requires too much effort to make work for me.
It helps that I’m an in-house video producer at a tech company and not working at a post house/on major sets anymore haha
The main current use case for this isn't people syncing their Dropboxes very fast, it's businesses and groups of users who can now all reliably each get gigabit+ speeds rather rather than slowing down when multiple people share a connection...
Sure, but I’m talking about my use case/the use case for most freelancers. We are shipping drives and transmitting footage constantly.
Hell fiber enabled me able to start charging people to send it quickly because it was cheaper than buying and shipping a drive (usually $80-$100). I just started going “I’ll have it to you in a few hours for $35” and call it a day.
What you need to keep in mind is that a lot of USB-C Ethernet adapters run over the USB3 lines - USB has a lot of CPU overhead which becomes visible at those speeds.
USB-C, with the right cable, connector and machine, can support Thunderbolt which is PCI-Express. If you go for a Thunderbolt adapter (and not a USB-based one, as above), you will get around this problem. You can also get one of those external GPU cases and put a desktop-grade PCIe network card in there and it should work provided your computer has the drivers for it (for Mac, I'd suggest trying an Intel card and hoping for the best).
I have this setup (eGPU case with a 10GbE NIC) and what you want for a Mac is an Aquantia AQC107 card, which I’ve managed to get a few of. Those have the same 10Gb chip as Apple uses so work perfect with no issue.
Chattanooga Tennessee, for those curious but not curious enough to read the article.
I'm intrigued by the city's push for this community co-op high speed internet - does anyone here have any experience with how the city has changed before/after the "gig-city" push? And did pandemic work from home change/accelerate things at all?
It started out to build a smart power grid, and I can say it did that pretty well. The network was able to automatically re-routr power to decrease outages substantially during storms without requiring crews to fix it.
Most people I know converted pretty quickly to EPB from Comcast, but apartments and HOAs still make contracts for exclusive Comcast wiring, even brand new HOAs that are being built.
My understanding is Comcast significantly improved in the Chattanooga area to compete, but I never went back and will never move to an area where I have to use Comcast again.
The best part about EPB was that they will service any house within their service area - for no crazy cost. I think initial setup was sometimes either completely waived or less than $100, no matter the house.
Also, the billing is nice - when they say $70 - it's $70 - no extra taxes or fees on top of that like most telecoms and utilities do.
As for actual changes - the city failed to truly capitalize on having a gig network. We wasted millions of dollars on a city-wide wifi system that was never turned on, due to political reasons (thr decision maker saw poor performance and shut it down after $300 million was already wasted putting up routers across the city.
The push I expected was smarter stoplights and traffic management, but that never happened. If anything it got worse.
I'm still stunned there isn't a unified service that can tell you WHERE FIBER IS AVAILABLE BY ADDRESS throughout the country or just in specific cities. Even Google Fiber's dedicated map is horrible and incredibly annoying to grok.
Literally, has anyone at this address had a FTTH connection installed? I'd even pay $12/mo or a one time fee to check. So it sounds like Chattanooga apts likely won't have support anytime soon?
I have been here for a while, in the earliest days of the gig there was a mini-rush of startups to Chatt. Low cost of living, relatively small/quiet city, access to basically all major southern cities (barring southern Florida) by either a quick flight or couple hour drive… makes sense that it just works out nicely.
I live about 20mi out from town, in a relatively rural area, and we still get excellent 1Gbps here. They wired as many drops as I asked for when we built here. It’s like a dream for a remote worker.
The city still has presence from some startups (I’m a bit out of the loop on who all is here), and I’ve known quite a few remote engineers who have lived in the area as well. Not sure how much this differs from an alternate universe where Comcast rules the land.
EPB is excellent all around. Service is incredibly consistent, it’s a bill I happily pay every month.
Had to move away for work -- but in my experience, the city has successfully reinvented itself and is now one of the most desirable places to live in the south.
Meanwhile I'm in the heart of a city and stuck with unreliable, overpriced, and low speed cable/dsl from Spectrum because I'm pretty sure my landlord is taking part of their payola scheme. I used to live in Chattanooga which makes me extra bitter about it.
Meanwhile here in my large California city the best service I can get at my apartment is 100/10. That's because it's negotiated by the landlord with some kind of corporate rate (it's not shared, we still have our own modem). It actually drops to something like 80 down at peak hours.
I've been shopping around for alternatives to Texas, and Chattanooga, TN just jumped to the top of my list.
I definitely couldn't take direct advantage of 25 gigabit/s service myself. But, I do enjoy the idea of it attracting certain types of businesses & people to an area and creating opportunities that might not be possible elsewhere.
Also, the fact that this sort of internet service is possible in an area gives me hope for the leadership.
I’ve had EPB for about 3 years now (since I moved to Chattanooga).
It’s fantastic. Every once in a while someone will post in the local Facebook group asking which ISP they should go with. It turns into pages and pages of EPB!, EPB!, EPB!
Mixed with comments about how much better it is than Comcast. Not only is the service better and cheaper, customer service is much much better.
People may not realize just how many barriers are put in the way of this kind of thing happening and at every level of government you have skilled and experienced actors trying to make sure it doesn't happen.
The obvious one is states pass laws to forbid municipal broadband. This should be illegal. I actually wonder if the Federal government could override this since this sort of thing usually falls within the purview of the FCC but I'm no lawyer. Comcat, AT&T and Verizon all lobby hard to make municipal broadband illegal, even when they won't connect homes they have a legal obligation to connect.
But imagine it is legal and you've started an ISP. Your problems have just begun.
How are you going to run cables (fiber or otherwise) to people's homes? Are you going to dig trenches? Well you need permits for that. You might dig up someone else's cables so existing providers get to delay that process.
Or are you going to string those cables up on poles? Well, who owns the poles? It might be an existing ISP. There might be laws that allows you to string cables on those poles. Maybe there isn't. Maybe you have to pay an exorbitant amount for the privilege. Maybe Comcast is the only one legally allowed to do cable work and they'll charge you a fortune for it and take forever (and probably screw it up).
Even if the utility poles are owned by the city or the county but that doesn't necessarily solve your problem either. You need permits to string up cables. You might have to file a permit application for each pole individually. Even when you get a permit, existing users may have to move their cables to make room. They may have 90 days to move it. If several need to move cables then may well take the maximum time in turn.
Either through trenches or poles you eventually have cables. Where do you run them to? As in, are you running short cables to substations and then trunk lines back to an exchange? Or are you running long cables back to an exchange directly? Each has its problems.
The long cables may make too thick bundles for poles or require more expensive trenching. The shorter cables may require a substation that the locals view as an eyesore. You'll need to acquire land for that and may face NIMBYist opposition.
What I learned through exposure to this was that building an ISP is hyperlocal and the incumbents are way better at it. For example, there may be a lot of limestone in the soil that may make trenching slow, difficult and expensive. You may want to use poles instead but that might not be an option so you're stuck with disposing of a lot of limestone.
Dicalimer: Xoogler and I worked on Google Fiber for a time.
This is just getting stupid. It's more marketing than anything else. It feels like the megahertz wars of the 90s. Your connection is SO MUCH more than just your local PHY rate. Let's start raising the bar. Show me you've optimized to reduce bufferbloat. Show me you care about RTT and not simply the cheapest pipe.
I'm also not sure they have the capacity to deliver true 25Gb to for many folks simultaneously.
If you are using so much bandwidth on a gigabit line that the buffers are actually used, I'd think you probably want a higher grade of service than home internet.
I'm not sure I agree. Many devices can saturate a 1Gb pipe these days. Even if it's for a brief period of time, your latency sensitive applications will suffer. Everyone should be using AQM.
In short no, but ironically the last time I read about rural 10G fiber the one guy they interviewed was a radiologist who wanted to work from home lmao. In short, I don't think he was too worried about dropping $1k on a router etc. Granted, if I remember correctly I think the ISP actually loaned him some enterprise routing gear.
hahaha found it... ironically it's a vice article [0]
Considering that 10G fiber gear is not very expensive at all (ie 28 port fiber switch - $1000) I’ve wondered what the hold-up is on 10Gbps SFP+ becoming a standard.
In Australia we’re shamefully paying $1150 per month for our business symmetrical 10G SFP+ (shaped to 1/1G), but a 1G 1000/40 connection is only $100/mo.
You still have to get transceivers for both ends of each and every connection which drives up the cost per port pretty dramatically.
Edit:
Scantly driving up the cost per port.
Say what? You've just added $1,500 (28 ports * 2 per port * $27) to the cost of a 28 port switch… at the low end.
Now I've no experience with fiber, but anecdotally I've come across a few complaints that the Ubiquiti stuff is picky about the transceivers you use. So if you go with a single manufacturer, Ubiquiti wants about $900 for their 28 port switch, but SFP+ transceivers start at $38. Their direct attach cables are cheaper but a 5m cable will still run you about $40. Compare that to *BaseT where I can grab a 3m Cat6 cable for under $10.
For an ISP the cost difference probably matters much less as they will be making money off of each connection (and likely saving on cooling costs). For a home user (e.g. an EPB customer) that's still pricey. An ER-X w/ a single SFP port runs about $80, using that SFP port will cost you.
Let's be generous and assume that we need far more reach than what is realistic in this scenario, take it at 10km between the subscriber demarc and the CO where these connections would presumably be terminated. A 10km LR optic is only $27 [0]. Amortize that over, say, three years and that's $0.75 per port, multiplied by two, so call it $1.50 per port/month over three years. Scantly driving up the cost per port. The physical fiber linking the two ends is more expensive (though you could depreciate and amortize over a longer period of time).
The 25G optics are barely that much more, LR/10km for $59 a pop. [1]
It's not that bad given the performance gain. Fiber patch cables are dirt cheap, FS.com optics are a great deal and also very affordable. Catch is to make sure you carefully buy hardware that accepts the proper optics, insane that it's somehow legal to lock down optics for specific vendors (I'm looking at you intel).
You're right. Definitely wouldn't consider the direct attach cables though for anything longer than 1m. Any serious deployment you'd want to be full optics.
Here in Europe it's mostly big ISPs that deploy PON. From what I understand it only makes sense with a large customer base. Small ISPs typically have P2P deployments and use standard switches.
This is incredible, I'm full remote and tired of paying high rent in austin (even after leaving NYC) which makes me want to consider TN a bit more. However, what kind of networking (ideally prosumer) hardware do you need to truly leverage this kind of a connection. Switches are cheap, I already have a few 10G fiber switches, however routers are another story, especially ones that are actually capable of keeping up with a 10G or 25G connection.
2G Google Fiber has been an absolute game changer - from experience I can also say it's better than 2x bonded Verizon Fios fiber connections. By far, the coolest feature of G Fiber is the online portal to request a static IP. Really hope static IP capability is also an option with this 10G service!
I usually buy used enterprise gear, and have been in a similar situation: switch ports are (relatively) cheap, anything that needs an actual CPU to do routing is spendy.
I've gone to Mikrotik for routing 10Gb in the home now. Excuse the horrible part number: the CCR2004-16G-2S+PC has worked out great for me. They're a bit hard to find right now, but I wouldn't hesitate to recommend them over whatever Ubiquiti is trying to sell in the space. For multi-gig routing, the Mikrotik RB5009 has also been a great piece of hardware.
25Gb is really getting into a different class of hardware entirely. If I needed to route at that speed, I'd seriously consider a Supermicro + Opnsense build.
Ohh I'm so jealous of this. I'm stuck with 15Mbps for ~$200/mo because of the monopoly here. Really hope I'll be able to get something better in the future. Hopefully this will put some pressure on other ISPs to get their stuff together :(
I used to have EPB some 6 years or so ago. They were by far the best ISP I have ever used. I was also privy to an organization that had access to their 10Gb connection.
A public internet service has been on their radar for about a decade now, and I proud to see they managed to get it working.
I don't really see the point - impressed that they offered it, but how many people truly need it?
I have 1G symmetric FTTH internet, the bottleneck is still the services I want to access at the other end - really makes no difference how much faster the pipe is, if the service you are using can't keep up.
Will some people benefit?? sure - a handful of power users doing massive uploads and downloads for commercial purposes... but the typical Netflix watcher or telecommuter really isn't going to benefit at all from from anything faster than about 100MBs up/down right now.
Like I said, I have 1G fiber connection, you know how long it takes me to watch a two hour Netflix movie at 1G speed? Two hours. You know how long that would take me on a 50Mbs connection? 2 hours. You know how long that would take me on a 25Gb connection? 2 hours.
That said, if I could buy a 25 Gb connection at a reasonable price I would.
20 years ago you'd have said the same thing about dsl to broadband, and if everyone had that same mentality, Netflix would still be mailing dvds and not offering 4k streaming. Innovation is good.
I think the difference is that 20 years Gbit ethernet was becoming standard issue on desktops while even 10Gbit ethernet is still pretty rare outside of the data centre and switch uplinks today.
25Gbit internet will be great for schools, libraries, and offices with multiple users but it's going to be a while before it becomes relevant for individual homes.
It's extremely relevant for putting a multifamily / condo / apartment building on the internet.
25Gbit is plenty fast enough to service dozens of units - more than twice the downlink local loop speed DOCSIS4 has, and Comcrap is still on DOCSIS 3.x.
10/40Gb hasn't been the standard in data centers in a number of years. My house is entirely 10Gb (save for wireless), in part because older enterprise gear at those speeds is so cheap.
25/50 and 100/400 have supplanted 10/40 in the data center, and 800Gb is here now.
Well, if cost of hardware on ISP side is fairly similar for 1G, 10G and 25Gbps, then why not offer also the fastest option just because you can?
> if the service you are using can't keep up.
Thats standard chicken and egg problem, why provide faster servers when ISPs are offering 1Gbps at most
IMO internet speed could be like electricity/power - like most people don't care how much power they can get from the grid, in the future we will have the same when it comes to the internet - it will almost always be fast enough for everything
In all seriousness, YES! File sizes are not getting smaller. If the bottleneck of data transference were my hardware, then we would live in a data utopia.
Imagine the size of files for the last 20 years and you could probably do a relatively close comparison for the next 20. I would say they have future-proofed their system for a long while.
I would also say that with the addition of IoT, there is going to be a LOT more casual traffic across the wire in people's homes/businesses.
> Imagine the size of files for the last 20 years and you could probably do a relatively close comparison for the next 20.
Is it? A some point usefulness plateaus.
I mean taking the streaming example, we can easily stream several 4K HDR streams within a 1G pipe, and 4K is basically retina-class unless you plan to project in a cinema, so anything above is virtually useless (just like the move to 24bit/192kHz is for listening).
The only way I can see this use case growing in size in any semi-useful way is by reducing compression ratio to eliminate artifacts.
Similarly picture size increase but I don't see people start sharing gigapixel pictures.
Maybe this could be an enabler of truly privacy respecting home self-hosting. Own your data, own your services. Maybe distributed storage like ipfs could benefit from that as well.
But size, I can only see us using more of it because we basically now have the ability to be inefficient, not because it's useful.
But hey, 20 years is basically impossible to project into with any reliability.
Not sure if we are going to see this in the future, but I'd love to see video streaming / content production move to 60fps. You're right that beyond 4k doesn't really give you anything but the improvement 24fps to 60 is absolutely massive.
Would also be nice to see video call apps upgrade from the absolute worst qualities to something nicer.
Doubling frame rate won't add bitrate like double, thanks to frame difference compression algorithm. Video call won't use much too. I'd say VR could be the next application that needs more bandwidth.
I don't disagree - in 10 years this will be different and so will my opinion - things we don't even know about will become common everyday necessities and may require those kind of speeds...but right now, I don't see it.
That's the maximum, right? I would imagine they have plans from sub-gigabit up to 25G. I know I'd absolutely love the option of 10G for a reasonable price (thousand(s)?). Heck, I'd be overjoyed to get symmetrical 100 Mbps, and I'd gladly pay a few hundred for it.
Meanwhile my Dad has 500/500 AT&T Fiber, and he pays half what I pay for 930/35 Charter Spectrum cable :(
At this point it's not so much about normal usage as the outliers.
For instance work from home got more popular. Say you have to transfer disk images. It's a lot more pleasant to transfer a couple TB at 25 Gb. You probably won't need that every day, but when you do need it, it's very nice not to have to wait hours or days for it to get through.
The way I see it is that by raising the bar like this, you're making more realistic speeds for average users a) more stable/common and b) less expensive.
More important is high quality high upload bandwidth not hidden behind CGNAT. We could actually move away from having to depend on Microsoft/Apple/Google/Amazon servers to host and deliver personal content.
> I have 1G symmetric FTTH internet, the bottleneck is still the services I want to access at the other end - really makes no difference how much faster the pipe is, if the service you are using can't keep up.
I can definitely hit this with one Steam download. If any more users want to download games, they can easily compete for 1G.
Netflix and other services have relatively low bitrate streams too. If I could get higher quality / higher bitrate streams, I’d prefer that over what most streaming services offer too.
I think it’s easy to think things are good enough without looking too deep into the details. You’re used to 1G. But a lot of people are used to a lot more and a lot less.
>>You’re used to 1G. But a lot of people are used to a lot more and a lot less.
Trust me, I am not used to it (at least not yet)- I have only had it for 5+ months - I have been WFH for more than 25 years and only had access to 3Mbs DSL before this.
Ive hit 5gbps at my dorm room, I had upto 10gbps only issue it was shared with the entire base and at the time only had 2 10gbps trunks out of the ISP. They later upgraded to 2 40gbps from NTT.
There are places like Switzerland where you actually can get a 25Gbps residental connection for a reasonable price. The actual cost lies in your backend infrastructure and the skills you need to use it (it's all enterprise equipment at that level). Everything from the routing to the switching to the servers that you're likely running on that connection can get real expensive.
So weird hearing about how amazing this speed is for a convention center when I have 25Gbps at home and mainly just use it to download movies and anime from Usenet.
Good for you? The majority of the world is on a fraction of that, much less the US. It just became easy to get 2 Gbps in some parts of the USA while others you max out at 25 Mbps down and 2 Mbps up. This is very much good news to everyone else.
The price is the same as for 1Gbps and 10Gbps (~60USD/month) - though the initial setup price differs: ~300USD for 25Gb/s vs <100USD for 1Gbps
The actual speed within network is indeed 25Gb/s to some fellow users running speedtest.net servers. Inside Switzerland it's maybe ~15Gb/s to some well connected servers in the area of Zurich, and ~10Gb/s across EU. Tested with speedtest.net and iperf.
I think in the EU the limit is sometimes not Init7 but the Speedtest.net servers. I ran a test to many of them and did get above 10Gbps but only to eu-25g.ookla.tierservers.com, which clearly has a 25Gbps NIC.
A bunch of other Speedtest.net servers were located in Amsterdam and took a very similar route via Cogent but none of the others broke the 10Gbps barrier.
It is for sure often Init7 though. Those peering relationships it built up have become a bit of a weakness because they're usually limited to a 10Gbps port.
I've been on Sonic 10G for almost a year now and LOVE it, but it's definitely been a sore spot to get things set up to expect those kinds of speeds - prosumer 10G switches are vastly more expensive per-port, wired consumer devices don't typically support 10GbaseT/SFP+, 2.5G/5G switches aren't broadly available and commodified (e.g. UniFi has limited offerings here) and WiFi6E is still mid-rollout (almost no client devices currently in market yet) meaning that clients can't reasonably expect >1gbps of goodput, even with a good link from a modern device to a modern AP. Then there's flakiness: when things get hot in my garage, my 10G switch just stops working. My Thunderbolt-to-10GbaseT adapter for my MacBook runs very hot. Lots of sharp edges here.
The more consumers are buying 10G equipment for their 10G home links, the faster prices will come down and reliability will come up - not just for 10G but also for 2.5G/5G equipment. Hats off to EPB for paving the way for 25G and keeping vendors diligent in mapping out the next generation of their equipment.