Author here. I co-founded Forte, which competes with Railway, so weigh this accordingly. Before this I built new services with our cofounders at AWS.
Railway's post mortem describes things that would have never made it past a whiteboarding session at AWS (multi tenant builds, overprivileged credentials, and a lack of separation between systems).
The post goes through each one. Railway's report was detailed enough to make this possible, and they deserve credit for owning up to the incident and impact. But hopefully this is a moment of reflection as well, because their report also exposes some downright silly engineering decisions that we were pretty surprised and disappointed to unpack.
Absolutely, that's why the report was disappointing to me -- this wasn't a zero day or unpreventable attack vector. It was well within their control to design proper isolation to begin with
Author here. I co-founded Forte, which competes with Railway, so weigh this accordingly. Before this I built new services with our cofounders at AWS.
Railway's post mortem describes things that would have never made it past a whiteboarding session at AWS (multi tenant builds, overprivileged credentials, and a lack of separation between systems).
The post goes through each one. Railway's report was detailed enough to make this possible, and they deserve credit for owning up to the incident and impact. But hopefully this is a moment of reflection as well, because their report also exposes some downright silly engineering decisions that we were pretty surprised and disappointed to unpack.
This all seemed somewhat preventable if different engineering decisions were made.
Absolutely, that's why the report was disappointing to me -- this wasn't a zero day or unpreventable attack vector. It was well within their control to design proper isolation to begin with
This tells me they don’t have internal security architecture experience, or best practice was intentionally ignored. Both is poor signal imho.
[flagged]