SYSTEM Cited by 1 source

Google Cloud Platform (GCP)

What it is

Google Cloud Platform (GCP) is Google's public-cloud infrastructure service — one of the three hyperscaler clouds alongside AWS and Azure. GCP surfaces on this wiki primarily as a named substrate that hosts services the wiki covers (Redpanda Cloud / BYOC on GCP, BigQuery, GCS, Pub/Sub, Cloud Spanner, etc.) and secondarily as a named outage subject (the 2025-06-12 global outage disclosed in Redpanda's retrospective).

Services represented on the wiki

• Google Cloud Storage (GCS) — object storage; GCP equivalent of Amazon S3.
• BigQuery — serverless analytical warehouse + Iceberg reader (named in the 2025-05-13 Redpanda BYOC Iceberg Topics post).
• Pub/Sub — managed messaging; named in the 2025-06-20 Redpanda retrospective as a customer-side data source that was also affected by the 2025-06-12 GCP outage.
• Private Service Connect (PSC) — private cross-project / cross-account VPC service reachability. Named in the 2025-06-20 Redpanda retrospective as the one optional critical-path dependency when enabled for a Redpanda Cloud GCP cluster.
• Cloud Spanner — globally-distributed relational database with external-consistency guarantees; named in the 2025-03-18 Redpanda CDC connectors post via the gcp_spanner_cdc connector.
• GoogleSQL — Spanner's SQL dialect.

Notable outage (2025-06-12)

On 2025-06-12, GCP experienced a global outage triggered by an automated quota update to their API management system. The incident is canonicalised on the wiki via Redpanda's 2025-06-20 retrospective (sources/2025-06-20-redpanda-behind-the-scenes-redpanda-clouds-response-to-the-gcp-outage).

Timeline reference points from Redpanda's vantage:

• 18:41 UTC — GCP TAM notification to Redpanda.
• 19:41 UTC — GCP identified triggering cause; mitigations applied.
• ~20:00 UTC+ (approximate) — us-central-1 regional outage, duration ~2 hours.
• 21:38 UTC — Redpanda considered the incident mitigated from its vantage.

The outage also cascaded to Cloudflare (Cloudflare's own post) via a cross-provider dependency. The GCP-side formal writeup is at status.cloud.google.com.

Canonical wiki framing: the outage is used as a teaching example in butterfly-effect in complex systems and systems-thinking-for- reliability, where an "automated quota update to their API management system" — a change with superficially narrow scope — cascaded into a global multi-service outage. Paired in Redpanda's framing with the 2024-07 Crowdstrike Falcon-update outage as a companion instance of the same structural gap.

Seen in

• sources/2025-06-20-redpanda-behind-the-scenes-redpanda-clouds-response-to-the-gcp-outage — Canonical wiki disclosure of GCP as outage substrate. 2025-06-20 Redpanda retrospective on how Redpanda Cloud GCP clusters remained stable through the 2025-06-12 global GCP outage. Names GCP Pub/Sub, GCS, VM-provisioning, and Private Service Connect as the substrate components that were degraded or in the critical path. Only one cluster out of "hundreds" in the GCP fleet was materially affected (a staging cluster in us-central-1 lost a node with ~2-hour replacement latency). The post also canonicalises Private Service Connect as "part of the critical path for reading and writing data to Redpanda" when enabled — a narrow-but- sharp caveat on GCP-network-path dependencies.
• sources/2025-05-13-redpanda-getting-started-with-iceberg-topics-on-redpanda-byoc — GCS + BigQuery tutorial for Redpanda BYOC Iceberg Topics.
• sources/2025-04-07-redpanda-251-iceberg-topics-now-generally-available — GCP as one of the three Iceberg-Topics GA clouds (alongside AWS and Azure).
• sources/2025-03-18-redpanda-3-powerful-connectors-for-real-time-change-data-capture — gcp_spanner_cdc connector for Cloud Spanner change streams.

Related

• systems/google-cloud-storage · systems/google-bigquery · systems/google-pub-sub · systems/gcp-private-service-connect · systems/cloud-spanner
• systems/aws-s3 — peer object-storage primitive on AWS.
• companies/cloudflare — 2025-06-12 cascade counterparty.
• systems/redpanda-cloud · systems/redpanda-byoc — canonical wiki GCP-hosted workloads.
• concepts/blast-radius · concepts/cell-based-architecture — the reliability framings Redpanda uses to contextualise the GCP outage.
• companies/google — the company behind GCP.