Yes, it is possible to point to existing Application Insights resource (which can be in any location).
If East US fails, other data centers should not be impacted. Application Insights should be able to log everything sent to it (assuming the app running in East US is still running).
Update: As Ivan mentioned, ingest endpoints are globally distributed. Though SDK will send data to nearest endpoint, 200 response means that that endpoint persisted data in the right data center.
It will take more time than for the same data center but the performance should not be impacted:
- There are direct and fast connections between data centers. So, for East US <=> West US (same region) the latency should be minimal.
- SDKs send data in background in batches, so the latency requirements for ingest calls are not strict for optimal performance (we have many high volume apps which upload telemetry data across regions {not even data centers}).