[avh02]

i'm not necessarily familiar with all the services (or the specific dependencies) on that list, but a lot of it will be a knock on effect from e.g: fb login not working as expected

People will very often see it as "service x with fb login is broken"

[florianmari]

I work in one of these companies and I can confirm that is a network outage and even our internal tools hosted in our datacenter (so no dependencies with fb) are down

[walrus01]

I sincerely hope that AS3356 hasn't broken a big chunk of the Internet again.

https://isc.sans.edu/forums/diary/CenturyLink+Outage+Causing...

https://www.zdnet.com/article/centurylink-outage-led-to-a-3-...

[dahfizz]

My bet is BGP. Its always BGP.

[oblio]

It's been a while since I studied routing, but have they fixed BGP security?

I remember reading about the protocol and thinking something like: this seems awfully "hippie-trust" levels of security.

Granted, this was 10 years ago so my recollection of BGP is super fuzzy :-D

[walrus01]

nope. proper and full implementation of BCP38, MANRS, full RPKI validation of all IP ranges and such is a long distance away.

within trusted groups of ISPs where the admins all know each other, in certain specific geographic regions, it's better than others.. I would say that the number of ISPs I could call "fully compliant" with current best practices, in the Portland-Seattle-Vancouver area, is higher than in many other parts of the world.

[ficklepickle]

How is shaw?

Anecdote: I was surprised to find they honour TTLs on their DNS service. Everyone else I have access to seems to return the new IP almost immediately, but shaw waits it out.

[curi0sity]

As someone interested in the internal workings of the internet, do you have any further reading on current best practices?

[dahfizz]

To be fair, BGP only runs between large network operators. The security / trust of routing agreements is mostly solved through business deals and contracts.

[walrus01]

Every time this happens I have a mental metaphorical image of somebody tripping over the $5 power strip that has the single point of failure route-server or route-reflector plugged into it.

[avh02]

I stand corrected :)

Any info on what network/operator/operator/etc is suffering the issue? (just so people know about it if it's public, i don't see anything on HN at least)

Thanks, and good luck!