Measure real latency to GCP region instead of using current location

[steren]

Instead of prompting for current location, we could do like https://gcping.com and send requests to Cloud Run services deployed on every GCP regions and measure latency.

The optimizer would then need to en adjusted to take latency into account in addition to geo coordinates.

Hosting Cloud Run services in many GCP regions comes with an additional maintenance burden and costs, and adds server side dependencies to this tool.

It is unclear if this is worth the investment.

[steren]

Google Cloud has released Network Intelligence Center, which exposes this info in the Cloud Console.

However, the data is not available as a machine readable public dataset yet.

[steren]

I have reached out to the team internally, but haven't heard back

[ccope]

Note, the metrics mentioned in Network Intelligence Center queries above are only generated for (some?) zones you're actively using, so you can't explore where you should deploy potential resources.