Migrate Network Volume so Serverless workers have better GPUs
Hello guys
I’ve been on serverless for about a year and it’s time to make sure my serverless endpoints provide the best experience on my web application. Currently I have 30 serverless workers and this week some of them have taken up to 7 minutes each to run a workflow which is way higher than the 15sec prediction time I expect. I have my serverless workers connected to my Network Storage in region CA-MTL-1 and unfortunately I only get 1 useful GPU type available for my workers (A40). I want the workers to access better GPUs and I think that means I need them to be connected to my network storage in different regions. Can I duplicate my network storage in several regions? Or have a central region and only some network storage folders are copied over? Who can help me with this?
2 Replies
At this time the console won't allow you to duplicate files between regions but you can deploy two pods in both regions and transfer the files or use our S3 API
Runpod Documentation
S3-compatible API - Runpod Documentation
Use Runpod's S3-compatible API to access and manage your network volumes.