Hello -
I’m designing an API for usage in a web application, and I decided to use Cloud Bigtable for fast response times. “Cold starts” are a known issue with bigtable, and I’m happy to accept that trade off. However, I feel as if I’ve been experiencing cold starts much longer than typical.
My research tells me most cold starts are 500ms or less, yet I’ve been consistently experiencing cold starts of ~10s.
I don’t anticipate this will be an issue in production, as the web app will be served with GKE in the same region as the Bigtable instance, but I do want to inquire and see if I am doing anything wrong.
For context, I am located in Montreal, Canada, and the Bigtable instance is located in South Carolina, USA. A colleague in Pittsburgh doesn’t experience these abnormally long cold starts.
Let me know if there’s any other information I can provide.
