R
Railway9mo ago
JonasJS

Pricing Questions

I'm looking for some clarification regarding Railway's pricing structure. I've noticed that the 'Hobby' plan costs $5 per month, and it includes a $5 resource credit 'gift.' I assume this $5 will just be deducted from my general usage costs. Furthermore, under the 'included' section, it mentions 8 GB RAM / 8 vCPU per Service. Is this the maximum configuration allowed on a service before needing to upgrade to a higher plan like 'Pro'? Or does every service get 8GB ram by default 😉 🤞 I'm also curious about whether the pricing is based solely on usage or if there's a per-service charge. For example, would I pay the same amount for two servers with 1GB of RAM each as I would for one server with 2GB of RAM? Any help in understanding these details would be greatly appreciated! Thanks in advance."
37 Replies
Percy
Percy9mo ago
Project ID: N/A
JonasJS
JonasJS9mo ago
N/A
Adam
Adam9mo ago
You are correct on all fronts. Each service can use up to 8vcpu and 8GB of RAM. There are no “configurations” on Railway like there are on GCP etc. Services will use what they need. You pay for what you use, so 2 services using 1GB of RAM each will cost the same as 1 service using 2GB of RAM
JonasJS
JonasJS9mo ago
Alright, thanks!
Fille
Fille9mo ago
I am really curious to how this works under the hood. Is there any podcast or youtube video or anything where this explained? Like, what happens if i spin upp 100 services but they only use 0.01 % of cpu. How are you able to support this behaviour? And how fast do I have access to the full 8 cpus? Instantly? Or will they slowly scale up over time ?
Floris
Floris9mo ago
it is most likely shared it's incredibly difficult to offer private CPU's at such good prices
Fille
Fille9mo ago
Yes, thats my guess too. So those 8 cpus that is supposed to be available, is probably not then? Thats why i would love to know whats going on exactly.
Floris
Floris9mo ago
cummulatively they will have enough to serve your max load but as not EVERYONE uses 100% load i barely use the vCPU at all 0.4 usually yet i pay so no one would be negatively impacted
Fille
Fille9mo ago
I really love the concept of being totally free to have both freedome to scale up to many cpus and still not have to pay for it when not using it. Would love to how reality looks.
Floris
Floris9mo ago
and because they can share it over they can offer u this crazy price i mean bro it's docker they just have a few big servers and they just throw those clusters full with docker images
Fille
Fille9mo ago
I provably just need to try it out
Floris
Floris9mo ago
in a very well managed fashion they have a free planbro i was on it for a month or so its really good im running roughly 8000 lines of python on not even 4$ usage a month
Fille
Fille9mo ago
Love it, but it just sounds too good to be true hehe
Floris
Floris9mo ago
No description
Fille
Fille9mo ago
I will definitly try
Floris
Floris9mo ago
just try the free one
Floris
Floris9mo ago
No description
Floris
Floris9mo ago
No description
Floris
Floris9mo ago
No description
Floris
Floris9mo ago
hope this helps
Fille
Fille9mo ago
What is you service doing? How much users? Load?
Floris
Floris9mo ago
depends usually 30 at peak concurrently algorithms for mostly financial data and an API for my frontend to talk to
Fille
Fille9mo ago
Ok kool thanks
Floris
Floris9mo ago
btw 1.37usd usage was a whole month
Floris
Floris9mo ago
No description
Floris
Floris9mo ago
im sure youll loveit what are you planning to run anyways?
Fille
Fille9mo ago
I already am hehe
Floris
Floris9mo ago
local databases ? i like the frontend
Fille
Fille9mo ago
Hmm i dont hink i can share links here ?? It got removed
Floris
Floris9mo ago
its fine i got it so what youre doing basicly
Fille
Fille9mo ago
Its a lot of services/modules parts/containers
Floris
Floris9mo ago
is high scale image classification ye? i mean bro you'll be fine wait i dm u
Fille
Fille9mo ago
Its bit more than that
Adam
Adam9mo ago
I read a few incorrect points in this convo so I just want to clarify. Each service has access to the full 8vCPU and 8GB ram. Effectively it’s as if each service is running on a machine that has 8vCPU and 8GB ram. There is always 8 of each available, but you’re only charged for the consumption I’m not sure how it works on the backend, but i’ve seen cases where a service that is using 0.4vCPU suddenly spikes to 8 when there is a DDOS attack. This indicates that your services aren’t categorized by their past usage. The full amount is always available @morpheus9071 will be helpful for you to read this too.
Floris
Floris9mo ago
i see i see thats nice to hear
Fille
Fille9mo ago
That is very impressive that you have managed to provide this to us users. I will hurry up and try this out!