Storage pricing question
Can someone explain storage pricing for stopped gpu pods please? When i stop a pod it says i'm going to be charged an hourly rate (something like .028?) for the storage unless I terminate the pod. The pricing page says it's .1/gb per month though. Which is it? The hourly rate seems prohibitively expensive, and i can't imagine that I'm supposed to recreate my pod every time i want to test something and then delete everything.
Solution:Jump to solution
Terminate the pod completely, instead use a network storage in a popular region.
Network storage are essentially like "external harddrives" you can imagine, which any pod can persist data too. Usually under /workspace.
For ex. if you launch a network storage under region A, you can launch two pods on that network storage, and they are sharing that drive. ...
5 Replies
Solution
Terminate the pod completely, instead use a network storage in a popular region.
Network storage are essentially like "external harddrives" you can imagine, which any pod can persist data too. Usually under /workspace.
For ex. if you launch a network storage under region A, you can launch two pods on that network storage, and they are sharing that drive.
The problem with network storage is that they can be for really large files be slow if ur reading directly from it (but you can just move it outside of /workspace onto the actual machine and will be faster + also it locks u into a region
But can be helpful to persist data on runpod
But yeah, I personally try not to just leave a Pod stopped
I either have files backed up on Firebase Filestorage, cause i wrote an easy wrapper to download / upload; and it gives me flexibility to just deploy and reset up there, or I have a network storage for my larger stuff in a region that is popular for the GPU type that I use the most often.
Runpod has CPU Pods, but unfortuinately they are only in RO region right now which still makes them limtied tbh.
So how do I actually know it's using that volume? I made the network volume, selected a size. Then i clicked deploy on the storage card and picked a gpu and all. Did it create it on my file storage or the pod storage?
so when u launched it said what is your volume mount location, if you have a network storage, it will usually by default be under /workspace
what this means is that anythign under /workspace will be persisted across all pods
for ex. if u launch another pod and look at /workspace ull see the storage is shared
What this also means is no longer do you "stop" a pod, you only have the option to directly terminate it
b/c there is no reason to stop it and pay for an idle reserve time
instead it is already being persisted to volume 🙂
also means u dont need to pray and hope that a GPU pod becomes avaliable to transfer ur data off b/c will always be safe
ok, i think i get it. thanks for the help