[martinald]

As has been pointed out on every Starlink thread, problem isn't latency or speed, it's contention. Each satellite only has 20gbit/sec per beam of capacity. It's likely even in small cities that will not be enough (it's only 1-3k users streaming Netflix for example). There is no way you can connect a city of even 100k people with 20gig/sec of capacity these days, never mind major metro areas many times that size.

Elon Musk has admitted this himself, and the problem gets worse over time (bandwidth requirements grow every year, but satellites capacity is fixed), so the 20gbps they have now will seem even more limited in 3-5 years when the satellites are becoming EoL.

This is not to say that Starlink is fundamentally flawed, it has great potential for rural access.

[lumberingjack]

I have friends on the Indiana Ohio border that are only offered ISP from a monopoly, ISP offers one plan $45 mo 60/40mbps. They have the first fiber optic Network ever laid in the country from adelphia cable but Time Warner Cable bought it and turned it off 20 years ago.

[sigmaprimus]

I just gave Starlink my address as per the email request they sent out today.

I'm currently on an 5Mb down 1Mb up LTE connection with a total data limit of 250Gb/Month for the low price of 100 bucks a month. I do have very low latency < 20ms, but last month my friends kid was over and managed to burn through 95 gigs playing and updating some game called PubG that ended up putting me over my limit and I got to pay an extra 50 bucks for the next tier of usage.

If I get the opportunity to try out Starlink I will be more than happy to do so as it's very frustrating being forced to use an ISP monopoly provider, especially when there is fiber available to properties less than 5 miles away from mine.

[34679]

You should really check out Visible. Unlimited 4G data, even on the hotspot.

$25 per month if you join a party, which takes about 5 minutes after searching for "Visible Party".

[Reason077]

I’ve been playing a bit of pubg recently, it’s included with the Google Stadia streaming gaming service. Fun game!

Stadia can easily burn through tens of GB per hour in bandwidth, and needs low latency, so it’s a good test of my 5G mobile connection. Good thing I have unlimited data!

[Dylan16807]

The median speed is close to 60/5 so despite them getting screwed on fiber that's not even bad for the US. And the price isn't extortionate, even!

[dx034]

60/40mbps isn't bad unless you want to share it with a lot of people (e.g. larger families). Sure, downloads won't be as fast as you'd like them to be but it's enough for 4k streaming so I don't think many people would notice the difference between this and a 200mbit line.

[tzs]

I recently did a substantial speed drop and barely noticed.

When I dropped cable TV and phone 4 months ago, I also dropped my internet speed from the 600 I had, figuring I’d start low and then raise it to what I need.

Options were 25, 100, 200, 300, 600, 1000, or 2000, for 50, 55, 70, 80, 90, 100, or 300 $/month.

I went with 100...and am still there 4 months later. The only times I notice it’s slower than my old 600 is OS and sometimes large app updates. (And for mobile devices, the difference is even less because my WiFi was a lot closer to 100 than 600).

It hasn’t affected working from home at all (I’ve been WFH for a couple years now so have plenty of prior experience with that at 600). I do occasionally need large downloads from work, but I’ve never gotten more than 20 Mbps out of our servers hosted at AWS.

I was worried a bit about upload. Nominally it was 15 for the 600 plan and is 5 for the 100 plan. Comcast over-performs in my area, so actual is about 19 for 600 and 7 for 100. (Download for those two is actually about 620-640 and 124). But the only large uploads I do aren’t very large. They are just large enough that on the old plan I would overlap them with other work rather than wait...so now I just get more other work done during the upload and still don’t really notice the wait. I get the same total work accomplished in the same time as before, just sequenced slightly different.

I thought I might be going to 300 if it turned out that streaming TV ran me into Comcast’s data cap and I needed to go unlimited. That would have been an extra $50/month, so 100 unlimited would be $105/month. But they also had an option where you could get unlimited for much less but only if you rented their modem. Total was $25/month, but it was only available on 300 or faster plans. A 300 plan with the modem/unlimited bundle would be $105/month, same as the 100 plan with my own modem and added unlimited. Still wouldn’t need 300, but if it is the same price as 100 why not?

But it turns out that even in my bingiest COVID lockdown month I only hit about 300 GB. Also, Comcast has since raised the cap from 1000 GB to 1229 GB, and dropped unlimited to $30/month, so I now anticipate staying at 100 indefinitely.

[martinald]

I agree, I have the option of 1000/1000 for £50/month or 150/150 for £25/month. Both unlimited. Despite pushing the connection very hard I ended up downgrading to 150. It's hard to get more than 200mbit on WiFi consistently, and unless you download onto an SSD with ethernet, downloads are constrained by hard drive speeds (especially over USB).

[noahtallen]

I have family on 6mbps down for somewhere around $70, so that's a pretty decent price for rural internet :p

[dillonmckay]

Both myself and another remote coworker, this is our situation, although it is closer to $60.

6Mbps/768kbps DSL

[samoa42]

thats brave!

make sure you have some decent AQM (qos) like cake or fq_codel. isp-gear probably has not. IQRouter was one of the first but there are others https://www.stoplagging.com/

[war1025]

> ISP offers one plan $45 mo 60/40mbps.

I pay $75/month for 30/30. So that sounds like a pretty good deal to me.

[Reason077]

> ”Each satellite only has 20gbit/sec per beam of capacity. It's likely even in small cities that will not be enough”

But you're not going to have just one satellite per city. Consider that Starlink already has approval for 12,000 satellites. And has applied for 30,000 more! At 20 Gbps each, that’s a lot of bandwidth.

When fully built out, any given point on the ground should have coverage from dozens (hundreds?) of satellites at any given time, spreading the load in regions of heavy demand.

[dx034]

But they have to move. Having 100 satellites serving NYC means you'd need a similarly fine mesh over the ocean.

I think they'll just make it more expensive than internet in cities. That way it's still the best option for rural areas without attracting too many customers in cities.

[Reason077]

Exactly. And in a dense city like NYC, many customers wouldn't have a clear line of sight to the sky anyway. You could install communal antennas/receivers on building roofs, but at that point you might as well just pay for fibre or terrestrial mobile broadband.

There may be niche customers in cities (data redundancy/resiliency etc), but in general, the biggest advantage will be for rural/remote customers who currently pay above the odds for substandard service.

[grecy]

> Each satellite only has 20gbit/sec per beam of capacity

Can this be upgraded in future versions of the sats? Is it a major undertaking requiring new ground station hardware, or something relatively easy that will increase with each "new" version of the sats?

[robszumski]

I imagine they will scale up the sats to have more beams, better tech, etc. They have a super heavy lift rocket coming online in the years between now and the current batch of sats being retired. What they will be replaced with will likely look nothing like today's model.

[e_y_]

To a certain extent they can increase total bandwidth by launching more satellites, although it's unlikely to ever make sense for urban areas where you could roll out fixed infrastructure much more cheaply. With satellite constellations, you can't just add satellites only where they're needed.

[wmf]

In 3-5 years each location may be covered by 10+ satellites, but that still won't be enough.

I think I saw a photo of a 5 Tbps Netflix POP recently.

[phkahler]

>> Each satellite only has 20gbit/sec per beam of capacity.

How many beams per sat? What's the minimum angle/distance between them?

[gnopgnip]

That isn't how the math works out for an ISP. You can serve on the order of 50k customers on a 20gbit/sec uplink.

[Reason077]

Not when you’re promising them each gigabit speeds, though. Which many ISPs do now days.

[gnopgnip]

On average, end users use about the same amount of bandwidth regardless of what speed plan they are on. Then planning is based on average and peak bandwidth

[fourthark]

Usually not a promise, usually “best effort”.

[sumedh]

Can adding more satellites solve that problem?

[martinald]

Sort of, but not really. A major metro probably has a terabit+ of peak demand these days. It's not going to be viable to have 50+ sats over all metro areas at once (that would probably suggest a spacing of a few km, requiring literally millions of satellites deployed globally).

The math just doesn't add up. It's a similar problem to why 4G/5G isn't a viable replacement for home internet service (yet) in many areas, apart from its much worse as it's easier to add new 5G base stations to densely cover cities than add new satellites.

[Reason077]

> "It's not going to be viable to have 50+ sats over all metro areas at once" ... "requiring literally millions of satellites"

That doesn't sound right. A satellite doesn't have to be directly overhead to be reachable. The coverage map below suggests that each satellite can cover a region 800-1000km in diameter, so with a fully deployed constellation of 12,000 to 42,000 satellites, there should easily be 50+ sats visible from any metro area. Perhaps even more than that.

https://satellitemap.space/indexA.html

[martinald]

I'm surprised each sat has such a large radius (especially when the constellation is fully deployed). This is actually a bigger problem than I thought if that map is correct.

This means satellites are going to cover a very large area. There might be 50 sats to cover the area from Boston down to Washington DC. That's 1tbit/sec of bandwidth to cover 100m potential customers.

Considering beta service afiak is using 600-1200 sats I imagine sat penetration will be significantly less. Considering cell towers between 4G and 5G can offer 20gbps _per tower_ these days I can't see 20gbps working across an entire state. Especially when it has been hyped as better than terrestrial internet. Even with a small number of users it will degrade rapidly at peak times when you get 1-3k users watching netflix.

[dx034]

It's also why Wifi, as amazing as new generations are, can't be the universal solution. Even well built installations in offices will quickly clog up during busy times. Wifi is important for devices where cables don't work, the same as satellite via internet or 5G only makes sense where fibre doesn't work.