relaying alerts between alertmanagers

[Samit Jain]

We're using a third-party cloud observability platform based on a standard Prometheus + AM architecture to provide metric observability for our on-premise systems. The platform provides standard HTTP APIs to ingest, read and monitor metrics, and exposes configuration for routing alerts in AM.

A key requirement is to integrate alerting with proprietary escalation systems that we deploy on-premise to manage alerts (e.g. Incident Management systems). Essentially, our goal to relay the alerts from the cloud platform to our self-hosted Alertmanager that we fully own and integrates well with our escalation systems through various integrations. 

 I was wondering if there is a recommended pattern for relaying alerts from the cloud AM to our own AM on-premise? An experimental approach which seemed feasible is by setting a low repeat_interval in cloud AM to send "raw" notifications to a web-hook proxy, which translates a 'notification' back to an 'alert' before sending it to on-premise AM over a secure connection. The on-premise AM on receiving the alerts will apply AM configuration and escalate downstream.

I was wondering if there are any issues or gotchas in AM with this approach that we may be overlooking? Is there a better approach to relay alerts between AMs?

Samit