[LibreQoS] monitoring twitter rtt looks useful

[LibreQoS] monitoring twitter rtt looks useful

[Dave Taht]

If we only knew which 100 ips it was.

https://blog.twitter.com/engineering/en_us/topics/infrastructure/2019/expand-the-edge

-- 
This song goes out to all the folk that thought Stadia would work:
https://www.linkedin.com/posts/dtaht_the-mushroom-song-activity-6981366665607352320-FXtz
Dave Täht CEO, TekLibre, LLC

Re: [LibreQoS] monitoring twitter rtt looks useful

[dan]

[-- Attachment #1.1: Type: text/plain, Size: 1495 bytes --]

It's pretty interesting and though they were 'surprised' at the results, as
someone who lives 'a lot' of hops from major datacenters this makes perfect
sense lol.
Here's a local Spectrum circuit to twitter:
[image: Screenshot 2022-11-13 at 3.45.00 PM.png]
and here's off one of my Lumen fiber circuits:
[image: Screenshot 2022-11-13 at 3.45.19 PM.png]

41ms is 'fine'... except this is about the best it gets here to a major
service that lives in tons of datacenters off a DIA 10G fiber circuit.
104.244.42.65/.1 are the two DNS records I get for twitter and have the
same numbers between them.

Also, not to 'brag' here but I'm rocking that solid 41ms and the furthest
edges of my network are 15ms out so my wISP service spanks Spectrum in
latency every time.  Also note that the 41ms is through a cake shaper at a
site that is passing about 1600x200Mbps right now.


On Sun, Nov 13, 2022 at 2:43 PM Dave Taht via LibreQoS <
libreqos@lists.bufferbloat.net> wrote:

> If we only knew which 100 ips it was.
>
>
> https://blog.twitter.com/engineering/en_us/topics/infrastructure/2019/expand-the-edge
>
> --
> This song goes out to all the folk that thought Stadia would work:
>
> https://www.linkedin.com/posts/dtaht_the-mushroom-song-activity-6981366665607352320-FXtz
> Dave Täht CEO, TekLibre, LLC
> _______________________________________________
> LibreQoS mailing list
> LibreQoS@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/libreqos
>

[-- Attachment #1.2: Type: text/html, Size: 2460 bytes --]

[-- Attachment #2: Screenshot 2022-11-13 at 3.45.00 PM.png --]
[-- Type: image/png, Size: 23231 bytes --]

[-- Attachment #3: Screenshot 2022-11-13 at 3.45.19 PM.png --]
[-- Type: image/png, Size: 20086 bytes --]

Re: [LibreQoS] monitoring twitter rtt looks useful

[Dave Taht]

[-- Attachment #1.1: Type: text/plain, Size: 2690 bytes --]

My specific point from TFA was that they had 100 anycasted IPs all over the
world,

the distance to their dns server shouldn't matter. remarkably *all* of my
tcpdump  traffic for the last 60 seconds was ipv6, the only ipv4s I can see
were 104.244.42.193 104.244.42.2
which are twitters. (really shocking they have no ipv6)

https://bgp.he.net/AS13414#_prefixes

really awful ping times over verizon 5g

root@balthazar:/home/d# ping 104.244.42.2
PING 104.244.42.2 (104.244.42.2) 56(84) bytes of data.
64 bytes from 104.244.42.2: icmp_seq=1 ttl=52 time=268 ms
64 bytes from 104.244.42.2: icmp_seq=2 ttl=52 time=89.5 ms
64 bytes from 104.244.42.2: icmp_seq=3 ttl=52 time=111 ms
64 bytes from 104.244.42.2: icmp_seq=4 ttl=52 time=134 ms
64 bytes from 104.244.42.2: icmp_seq=5 ttl=52 time=158 ms
64 bytes from 104.244.42.2: icmp_seq=6 ttl=52 time=59.6 ms


On Sun, Nov 13, 2022 at 2:50 PM dan <dandenson@gmail.com> wrote:

> It's pretty interesting and though they were 'surprised' at the results,
> as someone who lives 'a lot' of hops from major datacenters this makes
> perfect sense lol.
> Here's a local Spectrum circuit to twitter:
> [image: Screenshot 2022-11-13 at 3.45.00 PM.png]
> and here's off one of my Lumen fiber circuits:
> [image: Screenshot 2022-11-13 at 3.45.19 PM.png]
>
> 41ms is 'fine'... except this is about the best it gets here to a major
> service that lives in tons of datacenters off a DIA 10G fiber circuit.
> 104.244.42.65/.1 are the two DNS records I get for twitter and have the
> same numbers between them.
>
> Also, not to 'brag' here but I'm rocking that solid 41ms and the furthest
> edges of my network are 15ms out so my wISP service spanks Spectrum in
> latency every time.  Also note that the 41ms is through a cake shaper at a
> site that is passing about 1600x200Mbps right now.
>
>
> On Sun, Nov 13, 2022 at 2:43 PM Dave Taht via LibreQoS <
> libreqos@lists.bufferbloat.net> wrote:
>
>> If we only knew which 100 ips it was.
>>
>>
>> https://blog.twitter.com/engineering/en_us/topics/infrastructure/2019/expand-the-edge
>>
>> --
>> This song goes out to all the folk that thought Stadia would work:
>>
>> https://www.linkedin.com/posts/dtaht_the-mushroom-song-activity-6981366665607352320-FXtz
>> Dave Täht CEO, TekLibre, LLC
>> _______________________________________________
>> LibreQoS mailing list
>> LibreQoS@lists.bufferbloat.net
>> https://lists.bufferbloat.net/listinfo/libreqos
>>
>

-- 
This song goes out to all the folk that thought Stadia would work:
https://www.linkedin.com/posts/dtaht_the-mushroom-song-activity-6981366665607352320-FXtz
Dave Täht CEO, TekLibre, LLC

[-- Attachment #1.2: Type: text/html, Size: 4547 bytes --]

[-- Attachment #2: Screenshot 2022-11-13 at 3.45.00 PM.png --]
[-- Type: image/png, Size: 23231 bytes --]

[-- Attachment #3: Screenshot 2022-11-13 at 3.45.19 PM.png --]
[-- Type: image/png, Size: 20086 bytes --]