[eoswald]

Sorry, I have a hard time blaming Google for this, when Railway seems to be having increasing trouble keeping the platform stable. Something like this should NOT take down an ENTIRE service. There should be a backup when literally your business is about being the reliable backend. This just seems like poor planning to me.

[ryanisnan]

I don't quite know what you mean. Do you really expect Railway to use a multi-cloud architecture to host all of their client's projects? I suspect that would lead to a lower availability, all things considered.

[eoswald]

Well, in the same token, is it smart to base your ENTIRE architecture on a single cloud architecture? Isn't that why some of us build in fallbacks for AWS-hosted services? I mean, their enitre platform, both public and private facing, is running on the same thing. One error, one problem, takes out the entire service.

[irjustin]

Taking this at face value, this doesn't happen to AWS clients - at least I don't read about it here.

AWS may have data centers[0] go[1] down[2], but that's within expected bounds of standard ops.

[0] https://hooks.slack.com/services/TJ7HQS7FC/B0B5S7UTBJ4/PUHIC...

[1] https://www.aljazeera.com/news/2025/10/21/what-caused-amazon...

[2] https://netflixtechblog.com/lessons-netflix-learned-from-the...

[impulser_]

They literally own their own data centers. That's whats surprising about this. They are lying to their customers when they say they operate their own data center because obviously they don't if everyone's apps are down with GCP blocking their account.

[brookst]

Is it not possible that they own their own data center and have an unfortunate Google dependency?

Obviously a fiasco but I’m not prepared to call them liars when it could be an honest mistake.

[impulser_]

Then don't say your not a "Cloud on top of a cloud" provider.

They even made fun of cloud providers being down when AWS was down.

[Terr_]

I imagine there's also an important difference between:

1. We depend on X but could gracefully migrate to an alternate in a week if we really needed to.

2. All data is mirrored instantly so that we can do seamless fail-over in case X has its own outage.

[ryanisnan]

Oh, I see what you mean. Eh, it's possibly the same reason that AWS essentially goes down when us-east-1 goes down.

[cactusplant7374]

Disaster recovery is pretty expensive, right? Especially for their size.