I had once thought about architecting my own railway/Platform-as-a-service[0] and this was something that I used to really think quite a lot as Google cloud is quite finnicky and could ban you, so It doesn't inspire the greatest confidence in me that Railway hadn't really fixed it.
In my opinion, although not a one to one comparison but I would generally prefer to have my own hetzner/ovh or any other provider whom I can have a customized chat about it. For example, I once talked to upcloud out of curiosity and they give a 24-48 hours time for you to respond and don't have concept of auto-banning in that regards.
Anyways, didn't Railway have co-location itself too. I mean, I find it funny but railway's colocation blog was actually one that I still remember the vague feeling of the exact day that I read it and it definitely impacted me (positively?) but Its sad to see that railway didn't have proper plan and still relied on GCP and not colocation.
I wish to read a proper report from Railway team and I would be curious what they might do to prevent this from happening in future.
I also remember talking to a senior engineer at E2B which also runs on google cloud and they mentioned that they faced issues of such things too but I think that they were handling it by having a large team and better support but basically relying on something like GCP does give quite some headaches.
[0]: I still think about it and have parked my domain but I wish to actually create something in the more B2B focus precisely because of these bans and abuses preferably speaking. Another reason why I am not building it right now is that I don't feel comfortable providing a service and having people rely on it and hold it to greater standard and in the meanwhile I continue to learn more about servers and I wish to think more deeply before committing too! :-D
I had once thought about architecting my own railway/Platform-as-a-service[0] and this was something that I used to really think quite a lot as Google cloud is quite finnicky and could ban you, so It doesn't inspire the greatest confidence in me that Railway hadn't really fixed it.
In my opinion, although not a one to one comparison but I would generally prefer to have my own hetzner/ovh or any other provider whom I can have a customized chat about it. For example, I once talked to upcloud out of curiosity and they give a 24-48 hours time for you to respond and don't have concept of auto-banning in that regards.
Anyways, didn't Railway have co-location itself too. I mean, I find it funny but railway's colocation blog was actually one that I still remember the vague feeling of the exact day that I read it and it definitely impacted me (positively?) but Its sad to see that railway didn't have proper plan and still relied on GCP and not colocation.
I wish to read a proper report from Railway team and I would be curious what they might do to prevent this from happening in future.
I also remember talking to a senior engineer at E2B which also runs on google cloud and they mentioned that they faced issues of such things too but I think that they were handling it by having a large team and better support but basically relying on something like GCP does give quite some headaches.
[0]: I still think about it and have parked my domain but I wish to actually create something in the more B2B focus precisely because of these bans and abuses preferably speaking. Another reason why I am not building it right now is that I don't feel comfortable providing a service and having people rely on it and hold it to greater standard and in the meanwhile I continue to learn more about servers and I wish to think more deeply before committing too! :-D
Discussion: https://news.ycombinator.com/item?id=48200827