date:20201226





On 12/26/20 20:00, Tony Wicks wrote:


Actually the equipment vendor's build in this sort of situation is normally 
directly related to the availability of affordable chipsets from the likes of 
Broadcom. For example the chipset in my XGSPON router is a BCM6858. No vendor 
is going to spend money to produce a CPE that no one will buy. Once the likes 
of Broadcom produce an affordable solution then all the main vendors will roll 
out CPE in short order.


So by your logic, the chipset is the problem, otherwise, people will 
just keep buying more bandwidth for its own sake, even when they don't 
need it :-)?


Mark.

Re: 10g residential CPE





On 12/26/20 19:58, Michael Thomas wrote:



The thing is that the pandemic has changed the game on the ground: 
there is an actual feature differentiator to be had. But having dealt 
with the Linksys folks in the past I don't put out much hope that 
they'll take advantage of it. The software development side was a vast 
black hole where time stands still. It seems the entire industry is 
like that.


A jump from 10Mbps to 100Mbps is a differentiator.

A jump from 100Mbps to 1Gbps, even though more difficult, is also a 
differentiator.


A jump from 1Gbps to 10Gbps... yeah, as my Ugandan friend would say, 
"That's a hard paper".


One would ask, "What happened to all the Gbps in between :-)?"

Again, your issue isn't the bandwidth itself. Your issue is how people 
use devices, as well as the limitations of those devices themselves.


You can, pretty much, forget about much of the world using a wired 
device, going forward.


Mark.

Re: 10g residential CPE





On 12/26/20 17:55, Baldur Norddahl wrote:



Since a lot of ISP equipment only has tiny buffers you will generally 
be unable to get great downloads from sources far away.


This is true for any application, in general.

500ms vs. 1ms for download efficiency will always show you what they are 
made of, regardless of how ridiculous the buffers are.


The solution has always been to get content as close to the eyeballs as 
possible. The positive side-effect with this, also, is that downloads 
complete sooner, freeing up the line quicker.


Mark.

Re: 10g residential CPE





On 12/26/20 17:35, Mikael Abrahamsson wrote:



Perhaps there are some issues at other parts of the network that 
limits their speeds? I'm in Stockholm, Sweden, with plenty of local 
CDNs located just 1-3ms away from me.


The Swedish model (Stokab) is one to envy. If only other gubbermints had 
the political will to copy this. But alas.



Here the "truth" is that if you game, you need to have a wired 
connection to your gaming computer. All gamers "know" this.


I don't have experience with PS5 and perhaps what you're saying is 
true for that customer base. I'd say it's not true for Xbox or Steam 
customers as they see speed prominently displayed on the screen.


https://support.xbox.com/en-US/help/games-apps/troubleshooting/troubleshoot-slow-game-or-app-downloads-on-xbox-one 



"Go to My games & apps > Manage > Queue and note the download speed 
shown on the game or app that’s being installed. "


Very handy.

Never owned an Xbox, so didn't know this.

That said, as popular as gaming is, I'm not sure it represents the 
global FTTH demographic. Also, for day-to-day gaming, I believe the 
worry will mostly be about latency and packet loss, than raw throughput. 
Raw throughput will be key when you're downloading new games or updating 
old ones, and chances are this will happen less frequently than just 
regular playing.


Mark.

Re: Nashville

2020-12-26 Thread Tom Beecher

That's not exceptionally uncommon. I have seen the same thing before in
other sets of street view images.



On Fri, Dec 25, 2020 at 3:15 PM cosmo  wrote:

> I see the logo now :
>
> https://www.google.com/maps/@36.1640386,-86.7765438,3a,34.8y,283.85h,92.88t/data=!3m6!1e1!3m4!1sle30GenlolagNX2ldhGcwQ!2e0!7i16384!8i8192
>
> Most amazingly, in some, *but not all*, Google street view shots, the ATT
> logo is completely blurred out!
>
>
> https://www.google.com/maps/@36.1641167,-86.7765941,3a,75y,243.72h,94.77t/data=!3m6!1e1!3m4!1sgqbURHpzbT63-5U4Avv7cQ!2e0!7i16384!8i8192
>
> On Fri, Dec 25, 2020 at 11:58 AM Matt Brennan  wrote:
>
>> During their press conference, the Nashville Metro PD put the RV at 166
>> 2nd Ave N, which is across the street from the 185 2nd Ave N location.
>>
>> It's halfway up the block from 2nd & Commerce.
>>
>> On Fri, Dec 25, 2020 at 2:36 PM Andy Ringsmuth  wrote:
>>
>>> Definitely was not at that intersection.
>>>
>>> https://www.youtube.com/watch?v=5nygTJeu9fU
>>>
>>> That’s security camera footage from about 154 2nd Ave. The AT building
>>> is across the street to the right.
>>>
>>> Commerce Street is a block to the left.
>>>
>>> 
>>> Andy Ringsmuth
>>> 5609 Harding Drive
>>> Lincoln, NE 68521-5831
>>> (402) 304-0083
>>> a...@andyring.com
>>>
>>> “Better even die free, than to live slaves.” - Frederick Douglas, 1863
>>>
>>> > On Dec 25, 2020, at 1:26 PM, cosmo  wrote:
>>> >
>>> > The internet is buzzing with speculation about this. According to CNN
>>> the RV was at 2nd and Commerce st, which puts it 1-block away from the ATT
>>> building. If it were the target, I'd imagine they would have parked it
>>> closer.
>>> >
>>> >
>>> https://www.google.com/maps/search/2nd+and+congress+nashville/@36.1631367,-86.776487,18.42z
>>> >
>>> > On Fri, Dec 25, 2020 at 11:20 AM Andy Ringsmuth 
>>> wrote:
>>> > Certainly thankful no serious injuries or fatalities in this
>>> clusterblank.
>>> >
>>> > It seems the AT building at 185 2nd Ave N may have been a target,
>>> which would explain the timing (holiday morning when no one is out, as
>>> opposed to a holiday evening when there would be mass casualties). A little
>>> curious what that building has. Is it just a big co-lo place? Regional
>>> CLEC/ILEC?
>>> >
>>> > No earth-shattering revelations here. Admittedly just bored on a slow
>>> Christmas Day when my wife is at work (nurse) and kid is playing with a new
>>> tablet and I’m just watching the news trying to understand/figure out a
>>> little what and why.
>>> >
>>> > 
>>> > Andy Ringsmuth
>>> > 5609 Harding Drive
>>> > Lincoln, NE 68521-5831
>>> > (402) 304-0083
>>> > a...@andyring.com
>>> >
>>> > “Better even die free, than to live slaves.” - Frederick Douglas, 1863
>>> >
>>>
>>>

Re: [External] Re: 10g residential CPE

2020-12-26 Thread Jared Geiger

Are there 1G home routers that can do fq_codel in hardware versus the
general purpose CPU on the device? The only devices that I have that will
do a full 1G with it have active cooling fans.

It seems manufacturers need to meet that goal before we ask for 10G CPEs.

~Jared

On Sat, Dec 26, 2020 at 3:39 PM Mike Hammett  wrote:

> *nods* That leave delivering a better quality product to the rest of us.
>
> Ya know, peered well with whatever other networks may exist.  :-)
>
>
>
> -
> Mike Hammett
> Intelligent Computing Solutions 
> 
> 
> 
> 
> Midwest Internet Exchange 
> 
> 
> 
> The Brothers WISP 
> 
> 
> --
> *From: *"Mark Tinka" 
> *To: *nanog@nanog.org
> *Sent: *Friday, December 25, 2020 11:44:00 PM
>
> *Subject: *Re: [External] Re: 10g residential CPE
>
>
>
> On 12/25/20 23:04, Michael Thomas wrote:
>
> >
> > I mean, i understand the arm's race, but now it seems to be an arms
> > race for its own sake.
>
> It is, because it is hard to be different when all you know is to sell
> bandwidth.
>
> The next level of differentiation is being a fibre provider, and selling
> large amounts of bandwidth for less, and less, and less.
>
> It's a total lack of creativity in the infrastructure space, where the
> only goal is to maintain customers on the books.
>
> One of the mobile operators in South Africa, just last week, launched
> new data bundles for customers below a certain age. I mean, how many
> ways can you slice the selling of data because you can't be creative in
> other ways?
>
> Mark.
>
>

Re: Nashville

2020-12-26 Thread Javier J

I can confirm people in Alabama have outages as well. Reports from boots on
the ground in the Nashville area.

Comcast / Xfinity seem to be ok. Verizon seems to be ok also. Not sure the
logistics.

On Fri, Dec 25, 2020 at 3:38 PM Matt Hoppes <
mattli...@rivervalleyinternet.net> wrote:

> Can confirm internet service in Kentucky is being affected.
>
> On Dec 25, 2020, at 3:33 PM, Josh Baird  wrote:
>
> 
> I think the outage is a bit more widespread than "Nashville and
> surrounding areas."  Most (all?) of Kentucky is without AT cellular
> service right now.
>
> I can't say for sure of how many of AT's residential internet customers
> are affected, but reports on Twitter indicate it's a pretty significant
> chunk.  I have AT ASE/metro services here in Kentucky that do not appear
> to be affected at this time.
>
> On Fri, Dec 25, 2020 at 2:33 PM Sean Donelan  wrote:
>
>>
>> AT statement regarding the intentional explosion in Nashville TN
>>
>> "Service for some customers in Nashville and the surrounding areas may be
>> affected by damage to our facilities from the explosion this morning. We
>> are in contact with law enforcement and working as quickly and safely as
>> possible to restore service."
>>
>>
>> From local news reporting:
>>
>> A widespread internet outage was reported in Nashville hours after a
>> massive explosion downtown. AT internet and phone service was disrupted
>> in the area about 12 p.m. Friday.
>>
>> A handful of local police departments reported the outage was disrupting
>> 911 access, including some non-emergency lines, in their jurisdictions.
>>
>>
>>

Re: [External] Re: 10g residential CPE

*nods* That leave delivering a better quality product to the rest of us. 

Ya know, peered well with whatever other networks may exist. :-) 

- 
Mike Hammett 
Intelligent Computing Solutions 

Midwest Internet Exchange 

The Brothers WISP 

- Original Message -

From: "Mark Tinka"  
To: nanog@nanog.org 
Sent: Friday, December 25, 2020 11:44:00 PM 
Subject: Re: [External] Re: 10g residential CPE 

On 12/25/20 23:04, Michael Thomas wrote: 

> 
> I mean, i understand the arm's race, but now it seems to be an arms 
> race for its own sake. 

It is, because it is hard to be different when all you know is to sell 
bandwidth. 

The next level of differentiation is being a fibre provider, and selling 
large amounts of bandwidth for less, and less, and less. 

It's a total lack of creativity in the infrastructure space, where the 
only goal is to maintain customers on the books. 

One of the mobile operators in South Africa, just last week, launched 
new data bundles for customers below a certain age. I mean, how many 
ways can you slice the selling of data because you can't be creative in 
other ways? 

Mark.

Re: [External] Re: 10g residential CPE



On 12/26/20 3:28 PM, Mike Hammett wrote:
Use a router with FQ_CODEL and be amazed at how much you can get onto 
a pipe without any perceptible difference in the experience.




I did that, after a meltdown and yes it made a huge difference. I don't 
understand why CPE don't implement it by default.



Mike





-
Mike Hammett
Intelligent Computing Solutions 

Midwest Internet Exchange 

The Brothers WISP 


*From: *"Michael Thomas" 
*To: *nanog@nanog.org
*Sent: *Friday, December 25, 2020 1:27:39 PM
*Subject: *Re: [External] Re: 10g residential CPE


On 12/25/20 11:04 AM, Mark Tinka wrote:



On 12/25/20 20:10, Hunter Fuller via NANOG wrote:


It would meet some customers’ needs because multiple people
could use 1G of service at a time. I think it is interesting
to distinguish “>1G CPE” from “true 10G CPE” and I suspect
many / most customers are looking for the former.


Large upstream capacity has always been about aggregation of the
downstream.



Can I ask a really dumb question? Consider it an xmas present. I know 
this sounds like "nobody needs more than 640k", but how can household 
possibly need a gig let alone 10g? I'm still on 25mbs DSL, have cut 
the cord so all tv, etc is over the net. If I really cared and wanted 
4k I could probably upgrade to a 50mbs service and be fine. Admittedly 
it's just the two of us here, but throw in a couple of kids and I 
still don't see how ~100mbs isn't sufficient let alone 1 or 10G. Am I 
missing something really stupid?


Mike

Re: [External] Re: 10g residential CPE

Use a router with FQ_CODEL and be amazed at how much you can get onto a pipe 
without any perceptible difference in the experience. 




- 
Mike Hammett 
Intelligent Computing Solutions 

Midwest Internet Exchange 

The Brothers WISP 

- Original Message -

From: "Michael Thomas"  
To: nanog@nanog.org 
Sent: Friday, December 25, 2020 1:27:39 PM 
Subject: Re: [External] Re: 10g residential CPE 




On 12/25/20 11:04 AM, Mark Tinka wrote: 





On 12/25/20 20:10, Hunter Fuller via NANOG wrote: 








It would meet some customers’ needs because multiple people could use 1G of 
service at a time. I think it is interesting to distinguish “>1G CPE” from 
“true 10G CPE” and I suspect many / most customers are looking for the former. 



Large upstream capacity has always been about aggregation of the downstream. 







Can I ask a really dumb question? Consider it an xmas present. I know this 
sounds like "nobody needs more than 640k", but how can household possibly need 
a gig let alone 10g? I'm still on 25mbs DSL, have cut the cord so all tv, etc 
is over the net. If I really cared and wanted 4k I could probably upgrade to a 
50mbs service and be fine. Admittedly it's just the two of us here, but throw 
in a couple of kids and I still don't see how ~100mbs isn't sufficient let 
alone 1 or 10G. Am I missing something really stupid? 

Mike

Re: [External] Re: 10g residential CPE

Some WISPs I know moved customers from 20 megabit/s wireless to 500 megabit 
fiber. Total usage in that subdivision changed about 5%. 




- 
Mike Hammett 
Intelligent Computing Solutions 

Midwest Internet Exchange 

The Brothers WISP 

- Original Message -

From: "Michael Thomas"  
To: nanog@nanog.org 
Sent: Friday, December 25, 2020 1:27:39 PM 
Subject: Re: [External] Re: 10g residential CPE 




On 12/25/20 11:04 AM, Mark Tinka wrote: 





On 12/25/20 20:10, Hunter Fuller via NANOG wrote: 








It would meet some customers’ needs because multiple people could use 1G of 
service at a time. I think it is interesting to distinguish “>1G CPE” from 
“true 10G CPE” and I suspect many / most customers are looking for the former. 



Large upstream capacity has always been about aggregation of the downstream. 







Can I ask a really dumb question? Consider it an xmas present. I know this 
sounds like "nobody needs more than 640k", but how can household possibly need 
a gig let alone 10g? I'm still on 25mbs DSL, have cut the cord so all tv, etc 
is over the net. If I really cared and wanted 4k I could probably upgrade to a 
50mbs service and be fine. Admittedly it's just the two of us here, but throw 
in a couple of kids and I still don't see how ~100mbs isn't sufficient let 
alone 1 or 10G. Am I missing something really stupid? 

Mike

Re: [External] Re: 10g residential CPE

Ego. 
Ignorance. 




- 
Mike Hammett 
Intelligent Computing Solutions 

Midwest Internet Exchange 

The Brothers WISP 

- Original Message -

From: "Michael Thomas"  
To: nanog@nanog.org 
Sent: Friday, December 25, 2020 1:27:39 PM 
Subject: Re: [External] Re: 10g residential CPE 




On 12/25/20 11:04 AM, Mark Tinka wrote: 





On 12/25/20 20:10, Hunter Fuller via NANOG wrote: 








It would meet some customers’ needs because multiple people could use 1G of 
service at a time. I think it is interesting to distinguish “>1G CPE” from 
“true 10G CPE” and I suspect many / most customers are looking for the former. 



Large upstream capacity has always been about aggregation of the downstream. 







Can I ask a really dumb question? Consider it an xmas present. I know this 
sounds like "nobody needs more than 640k", but how can household possibly need 
a gig let alone 10g? I'm still on 25mbs DSL, have cut the cord so all tv, etc 
is over the net. If I really cared and wanted 4k I could probably upgrade to a 
50mbs service and be fine. Admittedly it's just the two of us here, but throw 
in a couple of kids and I still don't see how ~100mbs isn't sufficient let 
alone 1 or 10G. Am I missing something really stupid? 

Mike

Re: [External] Re: 10g residential CPE




On 12/26/20 1:13 PM, Valdis Klētnieks wrote:

On Sat, 26 Dec 2020 12:58:42 -0800, Michael Thomas said:

can go on for days. We have a generator because of this, but everybody
getting a generator in the middle of the Berkeley Hills would be
something of its own horror show, but it will probably come down to that.

Egads.

Especially if a lot of those generators are just bought at Home Depot and
hooked up to the house wiring without a proper cutover switch for the mains.




Yeah, it burned somebody's house to a crisp here last year around here. 
It certainly makes the case why leaving professionals in charge of power 
issues is the better idea. although with pg it's a tough call, my 
telco not so much.


Mike

Re: [External] Re: 10g residential CPE

2020-12-26 Thread Valdis Klētnieks

On Sat, 26 Dec 2020 12:58:42 -0800, Michael Thomas said:
> can go on for days. We have a generator because of this, but everybody
> getting a generator in the middle of the Berkeley Hills would be
> something of its own horror show, but it will probably come down to that.

Egads.

Especially if a lot of those generators are just bought at Home Depot and
hooked up to the house wiring without a proper cutover switch for the mains.




pgp3KpLpZtF4M.pgp
Description: PGP signature

Re: [External] Re: 10g residential CPE




On 12/26/20 12:44 PM, John Levine wrote:


In the 25 years since I've lived here the power has never been out as
long as a day so I think a four day battery will give me pretty good
reliability. I know my fiber is a straight shot to the CO since I'm
only four blocks away but as far as I can tell, unlike the HFC cable
plant next to it on the poles, their fiber system doesn't use any
powered repeaters.

Here in California the new reality is that multi-day outages are now 
common. The first few planned outages were 3-4 days, so that would be on 
the edge, especially if it's for gabby granny on the phone for 
hours.This all depends on the weather, and for snow related outages they 
can go on for days. We have a generator because of this, but everybody 
getting a generator in the middle of the Berkeley Hills would be 
something of its own horror show, but it will probably come down to that.


Mike

Re: [External] Re: 10g residential CPE

2020-12-26 Thread John Levine

In article <653758700.2275.1608968920711.javamail.zim...@baylink.com>,
Jay R. Ashworth  wrote:
>- Original Message -
>> From: "John Levine" 
>
>> They sure seem ready to take down the oopper. The installer was sad
>> when I told him to leave my six-pair copper cable alone even though
>> nothing is using it now.
>
>Sure; ILECs would *love* to deprovision their copper end networks.
>
>But that's not necessarily a great idea, societally; always-on dialtone
>(or, at least, dialtone with a much higher reliability than VoN) can be
>pretty important.  My LECs in Florida seem to manage five 9s pretty handily
>at the station set; betting FiOS isn't managing that. ...

My telco is a family run rural LEC with some quaint ideas. I asked the
installer who replaces the backup battery when it wears out.  "We do,
of course."  He seemed to think it was a silly question, was surprised
when I told him Verizon felt otherwise.

In the 25 years since I've lived here the power has never been out as
long as a day so I think a four day battery will give me pretty good
reliability. I know my fiber is a straight shot to the CO since I'm
only four blocks away but as far as I can tell, unlike the HFC cable
plant next to it on the poles, their fiber system doesn't use any
powered repeaters.

-- 
Regards,
John Levine, jo...@taugh.com, Primary Perpetrator of "The Internet for Dummies",
Please consider the environment before reading this e-mail. https://jl.ly

Re: 10g residential CPE



On 12/26/20 11:49 AM, Mel Beckman wrote:
The thing is that the pandemic has changed the game on the ground: 
there is an actual feature differentiator to be had. But having dealt 
with the Linksys folks in the past I don't put out much hope that 
they'll take advantage of it. The software development side was a 
vast black hole where time stands still. It seems the entire industry 
is like that.


Michael,

Even 100 Mbps Internet is fine for Zoom, as long as the uplink speed 
is at least 10 Mbps. The average zoom session requires 2 Mbps up and 
down, and even for the lavish six-screen executive sessions, 6 Mbps is 
plenty good. So arguing that 10 GbE is necessary because “pandemic has 
changed the game on the ground” is silly.


https://support.zoom.us/hc/en-us/articles/204003179-System-requirements-for-Zoom-Rooms#h_b48c2bfd-7da0-4290-aae8-784270d3ab3f 



So, sorry, 10 GbE is a hobbyists fantasy, not a marketable product. If 
hobbyists want 10GbE, let them pay for it like the rest of us, and let 
them play CoD from inside  freezing data center :)


I'm not saying anything about 10G, other than my initial query as to 
whether any residence could possibly need that much bandwidth. But 
buffer bloat is a problem with a lot of us still stuck on DSL with no 
prospect of anything better. It's not the bandwidth per se, it is how 
the bandwidth is consumed at home. Better queuing disciplines than tail 
drop with a gigantic queue could help zoom meetings a lot where 
bandwidth is more constrained. And regardless of bandwidth, huge queues 
are not good for real time traffic for anything. You'd think that gamers 
would be acutely aware of this and create a market for routers that 
cater to their hunger for less latency.


Mike

Re: 10g residential CPE

2020-12-26 Thread Mel Beckman

The thing is that the pandemic has changed the game on the ground: there is an 
actual feature differentiator to be had. But having dealt with the Linksys 
folks in the past I don't put out much hope that they'll take advantage of it. 
The software development side was a vast black hole where time stands still. It 
seems the entire industry is like that.

Michael,

Even 100 Mbps Internet is fine for Zoom, as long as the uplink speed is at 
least 10 Mbps. The average zoom session requires 2 Mbps up and down, and even 
for the lavish six-screen executive sessions, 6 Mbps is plenty good. So arguing 
that 10 GbE is necessary because “pandemic has changed the game on the ground” 
is silly.

https://support.zoom.us/hc/en-us/articles/204003179-System-requirements-for-Zoom-Rooms#h_b48c2bfd-7da0-4290-aae8-784270d3ab3f

So, sorry, 10 GbE is a hobbyists fantasy, not a marketable product. If 
hobbyists want 10GbE, let them pay for it like the rest of us, and let them 
play CoD from inside  freezing data center :)

 -mel

On Dec 26, 2020, at 11:12 AM, Filip Hruska  wrote:

 I wouldn't rely on these numbers too much, your testing methodology is flawed.
People don't expect RING nodes to be used as speedtest servers and so they are 
usually not connected to high speed networks.

Using a classical speedtest.net (Web or CLI) application would make much more 
sense, given the servers are actually connected to high speed Internet and are 
tuned to achieve such speeds - which is much more akin to how the most 
bandwidth demanding stuff (streaming, game downloads, system updates from CDNs) 
behaves.

It's certainly possible to get 1G+ over >10ms RTT connections single stream - 
the buffers are certainly not THAT small for it to be a problem - not to 
mention game distribution platforms do usually open multiple connections to 
maximise the bandwidth utilisation.

Re 85KB: that's just the initial window size, which will grow given tcp window 
scaling is enabled (default on modern Linux).

Filip

On 26 December 2020 19:14:13 CET, Baldur Norddahl  
wrote:

lør. 26. dec. 2020 18.55 skrev Mikael Abrahamsson 
mailto:swm...@swm.pp.se>>:
On Sat, 26 Dec 2020, Baldur Norddahl wrote:

> It is true there have been TCP improvements but you can very easily verify
> for yourself that it is very hard to get anywhere near 1 Gbps of actual
> transfer speed to destinations just 10 ms away. Try the nlnog ring network
> like this:
>
> gigabit@gigabit01:~$ iperf -c 
> netnod01.ring.nlnog.net
> 
> Client connecting to netnod01.ring.nlnog.net, 
> TCP port 5001
> TCP window size: 85.0 KByte (default)
> 
> [  3] local 185.24.168.23 port 50632 connected with 185.42.136.5 port 5001
> [ ID] Interval   Transfer Bandwidth
> [  3]  0.0-10.0 sec   452 MBytes   379 Mbits/sec

Why would you just use 85KB of TCP window size?

That's not the problem of buffering (or lack thereof) along the path, that
just not enough TCP window size for long-RTT high speed transfers.

That is just the starting window size. Also it is the default and I am not 
going to tune the connection because no such tuning will occur when you do your 
next far away download and wonder why it is so slow.

If you do the math you will realise that 379 Mbps at 10 ms is impossible with 
85 K window.

I demonstrated that it is about buffers by showing the same download from a 
server that paces the traffic indeed gets the full 930 Mbps with exactly the 
same settings, including starting window size, and the same path (Copenhagen to 
Stockholm).

Regards

Baldur

--
Sent from my Android device with K-9 Mail. Please excuse my brevity.

Re: 10g residential CPE

It was not meant to be a test as such, just a demonstration. Netnod to
Bahnhof is full speed and the third server is mine, so all three servers
can deliver at least 1G.

Finding a speedtest.net server at least 1000 km away that will show full
speed at 1G is hard. Namely because most such servers have at least 10G NIC
and that is not an advantage.

It is possible to get 1G at 10 ms, I did demonstrate that myself with the
test to Bahnhof. It is also possible to be limited at 30%. As the test to
Netnod shows.


lør. 26. dec. 2020 20.10 skrev Filip Hruska :

> I wouldn't rely on these numbers too much, your testing methodology is
> flawed.
> People don't expect RING nodes to be used as speedtest servers and so they
> are usually not connected to high speed networks.
>
> Using a classical speedtest.net (Web or CLI) application would make much
> more sense, given the servers are actually connected to high speed Internet
> and are tuned to achieve such speeds - which is much more akin to how the
> most bandwidth demanding stuff (streaming, game downloads, system updates
> from CDNs) behaves.
>
> It's certainly possible to get 1G+ over >10ms RTT connections single
> stream - the buffers are certainly not THAT small for it to be a problem -
> not to mention game distribution platforms do usually open multiple
> connections to maximise the bandwidth utilisation.
>
> Re 85KB: that's just the initial window size, which will grow given tcp
> window scaling is enabled (default on modern Linux).
>
> Filip
>
>
> On 26 December 2020 19:14:13 CET, Baldur Norddahl <
> baldur.nordd...@gmail.com> wrote:
>>
>>
>>
>> lør. 26. dec. 2020 18.55 skrev Mikael Abrahamsson :
>>
>>> On Sat, 26 Dec 2020, Baldur Norddahl wrote:
>>>
>>> > It is true there have been TCP improvements but you can very easily
>>> verify
>>> > for yourself that it is very hard to get anywhere near 1 Gbps of actual
>>> > transfer speed to destinations just 10 ms away. Try the nlnog ring
>>> network
>>> > like this:
>>> >
>>> > gigabit@gigabit01:~$ iperf -c netnod01.ring.nlnog.net
>>> > 
>>> > Client connecting to netnod01.ring.nlnog.net, TCP port 5001
>>> > TCP window size: 85.0 KByte (default)
>>> > 
>>> > [  3] local 185.24.168.23 port 50632 connected with 185.42.136.5 port
>>> 5001
>>> > [ ID] Interval   Transfer Bandwidth
>>> > [  3]  0.0-10.0 sec   452 MBytes   379 Mbits/sec
>>>
>>> Why would you just use 85KB of TCP window size?
>>>
>>> That's not the problem of buffering (or lack thereof) along the path,
>>> that
>>> just not enough TCP window size for long-RTT high speed transfers.
>>>
>>
>> That is just the starting window size. Also it is the default and I am
>> not going to tune the connection because no such tuning will occur when you
>> do your next far away download and wonder why it is so slow.
>>
>> If you do the math you will realise that 379 Mbps at 10 ms is impossible
>> with 85 K window.
>>
>> I demonstrated that it is about buffers by showing the same download from
>> a server that paces the traffic indeed gets the full 930 Mbps with exactly
>> the same settings, including starting window size, and the same path
>> (Copenhagen to Stockholm).
>>
>> Regards
>>
>> Baldur
>>
>>
>>
>>
> --
> Sent from my Android device with K-9 Mail. Please excuse my brevity.
>

Re: 10g residential CPE

2020-12-26 Thread Filip Hruska

I wouldn't rely on these numbers too much, your testing methodology is flawed.
People don't expect RING nodes to be used as speedtest servers and so they are 
usually not connected to high speed networks. 

Using a classical speedtest.net (Web or CLI) application would make much more 
sense, given the servers are actually connected to high speed Internet and are 
tuned to achieve such speeds - which is much more akin to how the most 
bandwidth demanding stuff (streaming, game downloads, system updates from CDNs) 
behaves. 

It's certainly possible to get 1G+ over >10ms RTT connections single stream - 
the buffers are certainly not THAT small for it to be a problem - not to 
mention game distribution platforms do usually open multiple connections to 
maximise the bandwidth utilisation. 

Re 85KB: that's just the initial window size, which will grow given tcp window 
scaling is enabled (default on modern Linux). 

Filip

On 26 December 2020 19:14:13 CET, Baldur Norddahl  
wrote:
>lør. 26. dec. 2020 18.55 skrev Mikael Abrahamsson :
>
>> On Sat, 26 Dec 2020, Baldur Norddahl wrote:
>>
>> > It is true there have been TCP improvements but you can very easily
>> verify
>> > for yourself that it is very hard to get anywhere near 1 Gbps of
>actual
>> > transfer speed to destinations just 10 ms away. Try the nlnog ring
>> network
>> > like this:
>> >
>> > gigabit@gigabit01:~$ iperf -c netnod01.ring.nlnog.net
>> > 
>> > Client connecting to netnod01.ring.nlnog.net, TCP port 5001
>> > TCP window size: 85.0 KByte (default)
>> > 
>> > [  3] local 185.24.168.23 port 50632 connected with 185.42.136.5
>port
>> 5001
>> > [ ID] Interval   Transfer Bandwidth
>> > [  3]  0.0-10.0 sec   452 MBytes   379 Mbits/sec
>>
>> Why would you just use 85KB of TCP window size?
>>
>> That's not the problem of buffering (or lack thereof) along the path,
>that
>> just not enough TCP window size for long-RTT high speed transfers.
>>
>
>That is just the starting window size. Also it is the default and I am
>not
>going to tune the connection because no such tuning will occur when you
>do
>your next far away download and wonder why it is so slow.
>
>If you do the math you will realise that 379 Mbps at 10 ms is
>impossible
>with 85 K window.
>
>I demonstrated that it is about buffers by showing the same download
>from a
>server that paces the traffic indeed gets the full 930 Mbps with
>exactly
>the same settings, including starting window size, and the same path
>(Copenhagen to Stockholm).
>
>Regards
>
>Baldur

-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.

Re: 10g residential CPE

On Sat, Dec 26, 2020 at 7:28 PM Mikael Abrahamsson  wrote:

> On Sat, 26 Dec 2020, Baldur Norddahl wrote:
>
> > I demonstrated that it is about buffers by showing the same download
> > from a server that paces the traffic indeed gets the full 930 Mbps with
> > exactly the same settings, including starting window size, and the same
> > path (Copenhagen to Stockholm).
>
> You demonstrated that it's about which TCP algorithm they use, probably.
>
>
All (virtual) machines used in the experiment are the same. Those are NLNOG
RING network managed machines all running the exact same Ubuntu 16.04.7 LTS.

If you have access to NLNOG RING or equivalent you should try the
experiment for yourself. You will find that as latency increases TCP speeds
goes down and it can not be explained by congestion. And you will find that
some servers have this effect much less than others and that those servers
usually have 1G network speed. The effect is the same no matter what time
of day you try it (ie. it is not congestion related).

Before you panic I will say I am not trying to advocate that we need more
buffers. We need smart buffers. Buffer bloat is bad but no buffers is also
bad. Your home made debloat solution will probably not be able to recover
the missing TCP performance that I am describing here. But if you could
have FQ Codel in the ISP switch that would probably do a lot.

Or we could have TCP with pacing and that will be widely deployed around
the same time as IPv6.

Regards

Baldur

Re: 10g residential CPE

2020-12-26 Thread Darin Steffl

Aaron,

One simple question. Why on earth would you offer free internet service?
How and why? Your site show 1 Gig symmetrical for free when you should be a
minimum of $65 per month to be competitive.

On Sat, Dec 26, 2020, 12:31 PM Aaron Wendel 
wrote:

> We run MikroTik RB4011s for residential speeds between 1G and 10G or just
> supply a media converter.  For residential 40G and 100G we just drop in
> Arista or Extreme switches.  SMBs are normally just a media converter or
> direct fiber handoff.
>
> https://mikrotik.com/product/rb4011igs_5hacq2hnd_in
>
> There are not a lot of options for good, off the shelf 10G CPE equipment.
> The handful of 10G residential customers we have seem to be happy with the
> tik.  The couple that don’t use it have rolled their own solution.
>
> Like anything, I’m sure once the major home broadband providers start to
> catch up with us smaller guys the vendors will catch up as well.
>
> https://www.kcfiber.com/residential
>
> Aaron
>
>
> On Dec 26, 2020, at 11:53 AM, Mel Beckman  wrote:
>
> 
>
> i really don't get what the problem is. it's like they're being
> deliberately obtuse.
>
>
> Michael,
>
> If vendors saw a 10GbE CPE market, they would serve it. Obviously they
> don’t see a market. Why don’t people insisting vendors build their hobby
> horse see that? It’s like they’re being deliberately obtuse :)
>
> -mel via cell
>
> On Dec 26, 2020, at 9:16 AM, Michael Thomas  wrote:
>
>
> 
>
> On 12/26/20 8:00 AM, Valdis Klētnieks wrote:
>
>
> Anybody got a feel for what percent of the third-party gear currently sold
> to
>
> consumers has sane bufferbloat support in 2020, when we've *known* that
>
> de-bufferbloated gear is a viable differentiatior if marketed right
> (consider the
>
> percent of families that have at least one gamer who cares)?
>
>
> I don't know percentages, but just trying to find cpe that support it in
> their specs is depressingly small. considering that they're all using linux
> and queuing discipline software is ages old, i really don't get what the
> problem is. it's like they're being deliberately obtuse. given all of the
> zoom'ing happening now you think that somebody would hit them with the
> clue-bat that this is a marketing opportunity.
>
>
> Mike
>
>
>

Re: 10g residential CPE

2020-12-26 Thread Aaron Wendel

We run MikroTik RB4011s for residential speeds between 1G and 10G or just 
supply a media converter.  For residential 40G and 100G we just drop in Arista 
or Extreme switches.  SMBs are normally just a media converter or direct fiber 
handoff.

https://mikrotik.com/product/rb4011igs_5hacq2hnd_in

There are not a lot of options for good, off the shelf 10G CPE equipment.  The 
handful of 10G residential customers we have seem to be happy with the tik.  
The couple that don’t use it have rolled their own solution.

Like anything, I’m sure once the major home broadband providers start to catch 
up with us smaller guys the vendors will catch up as well.

https://www.kcfiber.com/residential

Aaron


> On Dec 26, 2020, at 11:53 AM, Mel Beckman  wrote:
> 
> 
>> 
>> i really don't get what the problem is. it's like they're being deliberately 
>> obtuse. 
> 
> Michael,
> 
> If vendors saw a 10GbE CPE market, they would serve it. Obviously they don’t 
> see a market. Why don’t people insisting vendors build their hobby horse see 
> that? It’s like they’re being deliberately obtuse :)
> 
> -mel via cell
> 
>> On Dec 26, 2020, at 9:16 AM, Michael Thomas  wrote:
>> 
>> 
 On 12/26/20 8:00 AM, Valdis Klētnieks wrote:
>>> 
>>> Anybody got a feel for what percent of the third-party gear currently sold 
>>> to
>>> consumers has sane bufferbloat support in 2020, when we've *known* that
>>> de-bufferbloated gear is a viable differentiatior if marketed right 
>>> (consider the
>>> percent of families that have at least one gamer who cares)?
>>> 
>> I don't know percentages, but just trying to find cpe that support it in 
>> their specs is depressingly small. considering that they're all using linux 
>> and queuing discipline software is ages old, i really don't get what the 
>> problem is. it's like they're being deliberately obtuse. given all of the 
>> zoom'ing happening now you think that somebody would hit them with the 
>> clue-bat that this is a marketing opportunity.
>> 
>> Mike
>>

Re: 10g residential CPE


On Sat, 26 Dec 2020, Baldur Norddahl wrote:

I demonstrated that it is about buffers by showing the same download 
from a server that paces the traffic indeed gets the full 930 Mbps with 
exactly the same settings, including starting window size, and the same 
path (Copenhagen to Stockholm).


You demonstrated that it's about which TCP algorithm they use, probably.

They all respond very differently to increase in RTT vs loss.

https://arxiv.org/pdf/1903.03852.pdf

Generally the Internet doesn't need more buffers, it needs less. If you 
have only FIFO available, configure it to tail-drop at 10ms or so, to help 
your customers with what they really care about, interactive performance.


I debloat my 1000/1000 with bidir 900/900 FQ_CODEL to avoid my downloads 
affecting my interactive performance.


--
Mikael Abrahamssonemail: swm...@swm.pp.se

Re: 10g residential CPE




On 12/26/20 10:09 AM, Valdis Klētnieks wrote:

On Sat, 26 Dec 2020 17:50:28 +, Mel Beckman said:

If vendors saw a 10GbE CPE market, they would serve it. Obviously they donâ€™t
see a market. Why donâ€™t people insisting vendors build their hobby horse see
that? Itâ€™s like theyâ€™re being deliberately obtuse :)

The number of people that want a router that does 10GbE is vastly
outnumbered by the number of people that want a router that
makes their Zoom sessions not suck.

Admittedly, many of them don't realize they want that router, mostly
because most of them don't realize it's not difficult at all to build one
that does that.  But that's why companies have an advertising and marketing
team. :)


The marketing writes itself:

"Do you have to kick your kids of the network for company Zoom calls? 
You need this brand spanking new router!"


I've been trying to explain to friends that are now saddled with video 
calls all the time what the problem is, but it's really hard to point 
and say "buy this router". There are a few out there that feature it, 
but they're about $200 which is pretty spendy. Considering that this is 
just a OS module, your basic $50 router should be able to support it 
without any problem too.


Mike

Re: 10g residential CPE

lør. 26. dec. 2020 18.55 skrev Mikael Abrahamsson :

> On Sat, 26 Dec 2020, Baldur Norddahl wrote:
>
> > It is true there have been TCP improvements but you can very easily
> verify
> > for yourself that it is very hard to get anywhere near 1 Gbps of actual
> > transfer speed to destinations just 10 ms away. Try the nlnog ring
> network
> > like this:
> >
> > gigabit@gigabit01:~$ iperf -c netnod01.ring.nlnog.net
> > 
> > Client connecting to netnod01.ring.nlnog.net, TCP port 5001
> > TCP window size: 85.0 KByte (default)
> > 
> > [  3] local 185.24.168.23 port 50632 connected with 185.42.136.5 port
> 5001
> > [ ID] Interval   Transfer Bandwidth
> > [  3]  0.0-10.0 sec   452 MBytes   379 Mbits/sec
>
> Why would you just use 85KB of TCP window size?
>
> That's not the problem of buffering (or lack thereof) along the path, that
> just not enough TCP window size for long-RTT high speed transfers.
>

That is just the starting window size. Also it is the default and I am not
going to tune the connection because no such tuning will occur when you do
your next far away download and wonder why it is so slow.

If you do the math you will realise that 379 Mbps at 10 ms is impossible
with 85 K window.

I demonstrated that it is about buffers by showing the same download from a
server that paces the traffic indeed gets the full 930 Mbps with exactly
the same settings, including starting window size, and the same path
(Copenhagen to Stockholm).

Regards

Baldur

Re: 10g residential CPE

2020-12-26 Thread Valdis Klētnieks

On Sat, 26 Dec 2020 17:50:28 +, Mel Beckman said:
> If vendors saw a 10GbE CPE market, they would serve it. Obviously they donât
> see a market. Why donât people insisting vendors build their hobby horse see
> that? Itâs like theyâre being deliberately obtuse :)

The number of people that want a router that does 10GbE is vastly
outnumbered by the number of people that want a router that
makes their Zoom sessions not suck.

Admittedly, many of them don't realize they want that router, mostly
because most of them don't realize it's not difficult at all to build one
that does that.  But that's why companies have an advertising and marketing
team. :)


pgpveTfXzA_oP.pgp
Description: PGP signature

Re: 10g residential CPE




On 12/26/20 10:00 AM, Tony Wicks wrote:

Actually the equipment vendor's build in this sort of situation is normally 
directly related to the availability of affordable chipsets from the likes of 
Broadcom. For example the chipset in my XGSPON router is a BCM6858. No vendor 
is going to spend money to produce a CPE that no one will buy. Once the likes 
of Broadcom produce an affordable solution then all the main vendors will roll 
out CPE in short order.


Do they have no control of the linux kernel? This is purely OS kernel 
work and has nothing to do with underlying hardware.


Mike

RE: 10g residential CPE

2020-12-26 Thread Tony Wicks

Actually the equipment vendor's build in this sort of situation is normally 
directly related to the availability of affordable chipsets from the likes of 
Broadcom. For example the chipset in my XGSPON router is a BCM6858. No vendor 
is going to spend money to produce a CPE that no one will buy. Once the likes 
of Broadcom produce an affordable solution then all the main vendors will roll 
out CPE in short order.


>If vendors saw a 10GbE CPE market, they would serve it. Obviously they don’t 
>see a market. Why don’t people insisting vendors build their hobby horse see 
>that? It’s like they’re being >deliberately obtuse :)

Re: 10g residential CPE




On 12/26/20 9:50 AM, Mel Beckman wrote:

i really don't get what the problem is. it's like they're being deliberately 
obtuse.

Michael,

If vendors saw a 10GbE CPE market, they would serve it. Obviously they don’t 
see a market. Why don’t people insisting vendors build their hobby horse see 
that? It’s like they’re being deliberately obtuse :)

The thing is that the pandemic has changed the game on the ground: there 
is an actual feature differentiator to be had. But having dealt with the 
Linksys folks in the past I don't put out much hope that they'll take 
advantage of it. The software development side was a vast black hole 
where time stands still. It seems the entire industry is like that.


Mike

Re: 10g residential CPE


On Sat, 26 Dec 2020, Baldur Norddahl wrote:


It is true there have been TCP improvements but you can very easily verify
for yourself that it is very hard to get anywhere near 1 Gbps of actual
transfer speed to destinations just 10 ms away. Try the nlnog ring network
like this:

gigabit@gigabit01:~$ iperf -c netnod01.ring.nlnog.net

Client connecting to netnod01.ring.nlnog.net, TCP port 5001
TCP window size: 85.0 KByte (default)

[  3] local 185.24.168.23 port 50632 connected with 185.42.136.5 port 5001
[ ID] Interval   Transfer Bandwidth
[  3]  0.0-10.0 sec   452 MBytes   379 Mbits/sec


Why would you just use 85KB of TCP window size?

That's not the problem of buffering (or lack thereof) along the path, that 
just not enough TCP window size for long-RTT high speed transfers.


--
Mikael Abrahamssonemail: swm...@swm.pp.se

Re: 10g residential CPE

2020-12-26 Thread Mel Beckman

> i really don't get what the problem is. it's like they're being deliberately 
> obtuse. 

Michael,

If vendors saw a 10GbE CPE market, they would serve it. Obviously they don’t 
see a market. Why don’t people insisting vendors build their hobby horse see 
that? It’s like they’re being deliberately obtuse :)

-mel via cell

> On Dec 26, 2020, at 9:16 AM, Michael Thomas  wrote:
> 
> 
>> On 12/26/20 8:00 AM, Valdis Klētnieks wrote:
>> 
>> Anybody got a feel for what percent of the third-party gear currently sold to
>> consumers has sane bufferbloat support in 2020, when we've *known* that
>> de-bufferbloated gear is a viable differentiatior if marketed right 
>> (consider the
>> percent of families that have at least one gamer who cares)?
>> 
> I don't know percentages, but just trying to find cpe that support it in 
> their specs is depressingly small. considering that they're all using linux 
> and queuing discipline software is ages old, i really don't get what the 
> problem is. it's like they're being deliberately obtuse. given all of the 
> zoom'ing happening now you think that somebody would hit them with the 
> clue-bat that this is a marketing opportunity.
> 
> Mike
>

Re: 10g residential CPE

On Sat, Dec 26, 2020 at 5:41 PM Mikael Abrahamsson  wrote:

> On Sat, 26 Dec 2020, Baldur Norddahl wrote:
>
> > That is why. The RTT to the source can not be larger than the minimum
> > buffer size in the transport path. Otherwise the speed will start
> > decreasing.
>
> This is no longer correct. There has been lots of TCP innovation since
> this was true.
>
> Please stop repeating it.
>

It is true there have been TCP improvements but you can very easily verify
for yourself that it is very hard to get anywhere near 1 Gbps of actual
transfer speed to destinations just 10 ms away. Try the nlnog ring network
like this:

gigabit@gigabit01:~$ iperf -c netnod01.ring.nlnog.net

Client connecting to netnod01.ring.nlnog.net, TCP port 5001
TCP window size: 85.0 KByte (default)

[  3] local 185.24.168.23 port 50632 connected with 185.42.136.5 port 5001
[ ID] Interval   Transfer Bandwidth
[  3]  0.0-10.0 sec   452 MBytes   379 Mbits/sec

And that is a direct peer of ours.

In general you will have trouble with any server that has a NIC > 1G. If
you find a server that has a 1G NIC this happens instead:

gigabit@gigabit01:~$ iperf -c bahnhof01.ring.nlnog.net

Client connecting to bahnhof01.ring.nlnog.net, TCP port 5001
TCP window size: 85.0 KByte (default)

[  3] local 185.24.168.23 port 56412 connected with 195.178.185.171 port
5001
[ ID] Interval   Transfer Bandwidth
[  3]  0.0-10.0 sec  1.08 GBytes   930 Mbits/sec

Why? Because the 1G NIC server naturally will pace the traffic at maximum
1G and therefore not fill any buffers in the transfer path. The 10G servers
on the other hand WILL fill the buffers and experience packet loss.

Regards,

Baldur

Re: 10g residential CPE




On 12/26/20 8:00 AM, Valdis Klētnieks wrote:


Anybody got a feel for what percent of the third-party gear currently sold to
consumers has sane bufferbloat support in 2020, when we've *known* that
de-bufferbloated gear is a viable differentiatior if marketed right (consider 
the
percent of families that have at least one gamer who cares)?

I don't know percentages, but just trying to find cpe that support it in 
their specs is depressingly small. considering that they're all using 
linux and queuing discipline software is ages old, i really don't get 
what the problem is. it's like they're being deliberately obtuse. given 
all of the zoom'ing happening now you think that somebody would hit them 
with the clue-bat that this is a marketing opportunity.


Mike

RE: [External] Re: 10g residential CPE

2020-12-26 Thread Keith Medcalf

>If the operator wants to keep bufferbloat low you will not be able to
>utilise your 1 Gbps to that speed when downloading from distant servers.
>But with the same bufferbloat measured in milliseconds you will still
>have a 10x bigger buffer and thus 10x bigger bandwidth delay product.
>That translates to 10x the speed.

I should think that the speed were limited to some fraction of the speed of 
light being either the speed of signal propagation in copper or of photon 
travel in glass, and completely unrelated to bufferbloat or anything of that 
ilk.

1 Gps is a measure of volume, not of speed.  The speed is constant.

--
Be decisive.  Make a decision, right or wrong.  The road of life is paved with 
flat squirrels who could not make a decision.

Re: 10g residential CPE


On Sat, 26 Dec 2020, Baldur Norddahl wrote:

That is why. The RTT to the source can not be larger than the minimum 
buffer size in the transport path. Otherwise the speed will start 
decreasing.


This is no longer correct. There has been lots of TCP innovation since 
this was true.


Please stop repeating it.

--
Mikael Abrahamssonemail: swm...@swm.pp.se

Re: 10g residential CPE

2020-12-26 Thread Valdis Klētnieks

On Sat, 26 Dec 2020 00:32:49 -0500, b...@theworld.com said:

> I suppose that depends a lot on what the actual prices of a flat-rate
> 1gb vs a fully saturated 10gb. If it's $50 vs $100/mo perhaps some
> would say ok I'll risk the $50 overage, if it's $50 vs $500/mo maybe
> not.
>
> And today we have bandwidth-shaping in most any router/cpe (or could)
> so even with the 10gb/metered someone in the house with the password
> could rate-limit except when they needed it :-)

Note that the vast majority of users either use the ISP-provided CPE, or
something they picked up at Walmart or Best Buy.

This leads to an interesting economic incentive problem.  The ISP is obviously
not motivated to supply kit that can do bandwidth shaping on a metered drop.
Meanwhile, the providers of gear that gets sold at Walmart or Best Buy also
have no motivation to add it until enough ISPs are providing metered high-speed
service that "We can help prevent overage charges" becomes a viable market
differentiation.

Anybody got a feel for what percent of the third-party gear currently sold to
consumers has sane bufferbloat support in 2020, when we've *known* that
de-bufferbloated gear is a viable differentiatior if marketed right (consider 
the
percent of families that have at least one gamer who cares)?



pgpy2ZpJwQdKA.pgp
Description: PGP signature

Re: 10g residential CPE

lør. 26. dec. 2020 16.35 skrev Mikael Abrahamsson via NANOG :

>
>
>
> Perhaps there are some issues at other parts of the network that limits
> their speeds? I'm in Stockholm, Sweden, with plenty of local CDNs located
> just 1-3ms away from me.
>

That is why. The RTT to the source can not be larger than the minimum
buffer size in the transport path. Otherwise the speed will start
decreasing.

Since a lot of ISP equipment only has tiny buffers you will generally be
unable to get great downloads from sources far away.

The other option is huge buffers which gives you better speed but with
buffer bloat issues.

Regards

Re: 10g residential CPE

On Sat, 26 Dec 2020, Mark Tinka wrote:

My experience with customers who've bought 1Gbps FTTH service is that on a
good day, they may see 500Mbps. On average, they'll live somewhere between
180Mbps - 350Mbps, with a random spot-check. It's alright for providers who
offer this to let their NOC's handle the problem, because most users are
connected to the Internet wirelessly, using devices that do not require more
than a couple of Mbps of bandwidth at a time.

Wired devices such as gaming consoles won't tell you anything more than how
long it will take a download to complete. So you are not probably going to
work out whether the PS5 is running at 1Gbps or 230Mbps, as long as your
psyche is happy with the service you are buying from your provider.

Steam and Microsoft will say download speed. I regularily see 100MB/s or
more.

Perhaps there are some issues at other parts of the network that limits
their speeds? I'm in Stockholm, Sweden, with plenty of local CDNs located
just 1-3ms away from me.

Here the "truth" is that if you game, you need to have a wired connection
to your gaming computer. All gamers "know" this.

I don't have experience with PS5 and perhaps what you're saying is true
for that customer base. I'd say it's not true for Xbox or Steam customers
as they see speed prominently displayed on the screen.

https://support.xbox.com/en-US/help/games-apps/troubleshooting/troubleshoot-slow-game-or-app-downloads-on-xbox-one

"Go to My games & apps > Manage > Queue and note the download speed shown
on the game or app that’s being installed. "

--
Mikael Abrahamssonemail: swm...@swm.pp.se

Re: 10g residential CPE





On 12/26/20 16:38, Mikael Abrahamsson wrote:



Considering my PC often saturates my 1000/1000 Internet access when 
downloading, I don't see why the 1GE NIC on PS5 wouldn't be the 
bottleneck if it's sitting on higher speed Internet access.


My experience with customers who've bought 1Gbps FTTH service is that on 
a good day, they may see 500Mbps. On average, they'll live somewhere 
between 180Mbps - 350Mbps, with a random spot-check. It's alright for 
providers who offer this to let their NOC's handle the problem, because 
most users are connected to the Internet wirelessly, using devices that 
do not require more than a couple of Mbps of bandwidth at a time.


Wired devices such as gaming consoles won't tell you anything more than 
how long it will take a download to complete. So you are not probably 
going to work out whether the PS5 is running at 1Gbps or 230Mbps, as 
long as your psyche is happy with the service you are buying from your 
provider.


Mark.

Re: 10g residential CPE


On Sat, 26 Dec 2020, Mark Tinka wrote:

No one argued that Sony could build a half-decent console. Wired via 
Ethernet, that's unlikely to be the bottleneck.


Considering my PC often saturates my 1000/1000 Internet access when 
downloading, I don't see why the 1GE NIC on PS5 wouldn't be the bottleneck 
if it's sitting on higher speed Internet access.


--
Mikael Abrahamssonemail: swm...@swm.pp.se

Re: 10g residential CPE





On 12/26/20 15:45, Niels Bakker wrote:



Why wouldn't it go even faster, assuming it got fitted out with a 
faster network controller than what they shipped with?  The storage 
system in the PS5 as sold can transfer at 5 GB/sec and the APUs have 
the regular set of crypto acceleration instructions.
https://www.theverge.com/2020/11/5/21551165/sony-ps5-playstation-5-no-m2-ssd-expansion-launch 

https://www.tweaktown.com/news/71340/understanding-the-ps5s-ssd-deep-dive-into-next-gen-storage-tech/ 



No one argued that Sony could build a half-decent console. Wired via 
Ethernet, that's unlikely to be the bottleneck.


Mark.

Re: 10g residential CPE

2020-12-26 Thread Niels Bakker


* mark.ti...@seacom.com (Mark Tinka) [Sat 26 Dec 2020, 06:48 CET]:

On 12/25/20 23:22, Niels Bakker wrote:

Download times:-
180GB at 100 Mbps: 4 hours
180GB at 1000 Mbps: 23 minutes


For a number of reasons, highly unlikely your console will pull at 
1Gbps, but yes, it would certainly pull quicker than 100Mbps :-).


Why wouldn't it go even faster, assuming it got fitted out with a 
faster network controller than what they shipped with?  The storage 
system in the PS5 as sold can transfer at 5 GB/sec and the APUs have 
the regular set of crypto acceleration instructions.

https://www.theverge.com/2020/11/5/21551165/sony-ps5-playstation-5-no-m2-ssd-expansion-launch
https://www.tweaktown.com/news/71340/understanding-the-ps5s-ssd-deep-dive-into-next-gen-storage-tech/


-- Niels.

Re: [External] Re: 10g residential CPE





On 12/26/20 13:41, Nuno Vieira wrote:

Once upon a time a wise main said “Who in their right mind would ever 
need more than 640k of ram?”


While everyone will take a chance at using this line at some point in a 
computing career, it's somewhat disingenuous to compare (or equate) the 
640KB of RAM to anything today. There is always bare minimums we must 
hit before any additional infrastructure does not yield any further 
value. In economics, I believe it's called the Law of Diminishing Returns.


640KB may not be sufficient for any personal device today. However, what 
would you do with 256GB of RAM on your computer that you can't do with 
64GB, in a home/domestic setting, as an average user?


Yes, 1Mbps of bandwidth is likely too small for today's Internet, but 
what would you do with 10Gbps that you can't with, say, 750Mbps?


Maybe we shall have VR on our devices which will make Zoom much more 
pleasant to use, but for as long as you don't see any 10Gbps CPE or 
XG-PON roll-outs happening en masse, that should signal something to you.


There was a time when the number of megapixels on a phone or camera was 
the most important thing. That doesn't matter anymore, to the point of 
actually more megapixels returning poorer image quality. Just compare a 
photo taken in WhatsApp vs. a photo taken in Apple's native Camera app, 
same device. The smarts aren't in the hardware anymore.


Similarly, while 10Gbps to the home may have sounded completely 
noteworthy in 2005, surprisingly, perhaps less so in 2020. Likely 
because the way the Internet works, how data is delivered, how it is 
stored, the role of the device, e.t.c., all make the case for either 
more bandwidth (on the provider/backbone side) or less bandwidth (on the 
consumer side).


Mark.

Re: [External] Re: 10g residential CPE





On 12/26/20 13:41, Nuno Vieira wrote:


Once upon a time a wise main said “Who in their right mind would ever need more 
than 640k of ram?”


While everyone will take a chance at using this line at some point in a 
computing career, it's somewhat disingenuous to compare (or equate) the 
640KB of RAM to anything today. There is always bare minimums we must 
hit before any additional infrastructure does not yield any further 
value. In economics, I believe it's called the Law of Diminishing Returns.


640KB may not be sufficient for any personal device today. However, what 
would you do with 256GB of RAM on your computer that you can't do with 
64GB, in a home/domestic setting, as an average user?


Yes, 1Mbps of bandwidth is likely too small for today's Internet, but 
what would you do with 10Gbps that you can't with, say, 750Mbps?


Maybe we shall have VR on our devices which will make Zoom much more 
pleasant to use, but for as long as you don't see any 10Gbps CPE or 
XG-PON roll-outs happening en masse, that should signal something to you.


There was a time when the number of megapixels on a phone or camera was 
the most important thing. That doesn't matter anymore, to the point of 
actually more megapixels returning poorer image quality. Just compare a 
photo taken in WhatsApp vs. a photo taken in Apple's native Camera app, 
same device. The smarts aren't in the hardware anymore.


Similarly, while 10Gbps to the home may have sounded completely 
noteworthy in 2005, surprisingly, perhaps less so in 2020. Likely 
because the way the Internet works, how data is delivered, how it is 
stored, the roll of the device, e.t.c., all make the case for either 
more bandwidth (on the provider/backbone side) or less bandwidth (on the 
consumer side).


Mark.

Re: [External] Re: 10g residential CPE


On Fri, 25 Dec 2020, Chris Adams wrote:

Queueing doesn't get me my next game in time to play it tonight.  I've 
always seen general queueing as a work-around for "not enough bandwidth 
and can't add more"... but when more is available, why not just use 
more?


I de-bloat my 1000/1000 with FQ_CODEL. It's worthwhile because even 
1000/1000 can see RTT spikes of tens of milliseconds otherwise.


Bandwidth doesn't solve queuing and queuing doesn't solve bandwidth. 
They're both needed.


--
Mikael Abrahamssonemail: swm...@swm.pp.se

Re: [External] Re: 10g residential CPE