This document outlines Google Cloud's approach to communicating incidents,
including the roles of the following primary communication channels:
Personalized Service Health (PSH) and the public Google Cloud Service Health (CSH)
dashboard.

## Where to look for incident communications

Google Cloud provides two channels for incident communication, each with a
different purpose:

- **Personalized Service Health (PSH):** This is your primary source for service
  disruptions relevant to you. It provides a personalized view tailored to
  the specific Google Cloud products you use. We recommend integrating
  Personalized Service Health into your incident response process as a critical
  corroboration signal.

  To avoid paging an on-call engineer for every Personalized Service Health event,
  integrate Personalized Service Health incident visibility into your team's
  dashboards and tools. This practice helps operators quickly determine
  if a suspected issue is related to a Google Cloud service disruption.
  [Learn more about PSH](https://docs.cloud.google.com/service-health/docs/overview.md.txt).
- **Google Cloud Service Health (CSH):**
  This is Google Cloud's public-facing status page, available at
  [status.cloud.google.com](https://status.cloud.google.com/). Google Cloud
  Service Health requires no login and serves as an at-a-glance health check
  for the entire platform, and is used to communicate broad severe incidents or when
  PSH itself is unavailable.

![A diagram comparing PSH to CSH communication channels. The diagram shows emerging incidents,
confirmed incidents, and broad severe incidents going to Personalized Service Health.
The diagram shows broad severe incidents going to Cloud Service Health. Listed under
Personalized Service Health, is Dashboard, API, and Cloud Logging. Listed
under Cloud Service Health is Dashboard and RSS Feed.](https://docs.cloud.google.com/service-health/images/comparison-psh-csh.png)

## Our disclosure strategy and recommendations

Deciding what to share, and where, is not an arbitrary process. It is a formal,
systematic discipline based primarily on an incident's scope. The intent of these
channels is to provide maximum visibility for Broad Severe Incidents while reducing
the noise of incidents that are irrelevant.

- **For Broad Severe Incidents:** Broad-scoped incidents---those impacting a large
  percentage of projects or widespread across multiple regions---are communicated
  using Google Cloud Service Health (CSH). These incidents are also communicated
  to affected customers using Personalized Service Health. This helps ensure the message
  reaches the widest possible audience for the most critical events.

- **For other Confirmed Incidents:** For issues with a more limited scope,
  such as those impacting a single location, zone, or a smaller subset of
  projects, we will communicate relevant incidents to customers using
  Personalized Service Health. We aim to be comprehensively transparent, which means
  Service Health makes available all potentially relevant events for
  your services. If you need a more focused event feed, Service Health
  offers tools to filter and fine-tune the events that are passed through your
  alerts and automated workflows.

  - **Recommendation:** Configure alerts to only focus on the most critical Google Cloud services and locations, or to trigger only on events with a relevance of "Related" or "Impacted." See examples of [how to filter and fine-tune alerts](https://docs.cloud.google.com/service-health/docs/example-alerting-policies.md.txt).

### Fallback Strategies for Personalized Service Health unavailability

Personalized Service Health depends on core services, such as Identity and Access Management for
authentication. In a severe, widespread disruption, the very services you need
to sign in might be affected.

We recommend the following fallback strategy:

- **For manual processes:** Your runbooks should direct operators to the
  Google Cloud Service Health dashboard at [status.cloud.google.com](https://status.cloud.google.com/)
  in the event they can't access the Personalized Service Health dashboard.

- **For automated systems:** Use the [**Service Health Status API**](https://docs.cloud.google.com/service-health/docs/service-health-disruption.md.txt)
  to programmatically detect if Personalized Service Health is having a problem.
  If it is, your systems can then fall back to ingesting the public
  [**CSH RSS Feed**](https://status.cloud.google.com/en/feed.atom) for continued
  programmatic updates.