SYSTEM Cited by 3 sources

Google Cloud Storage

What it is

Google Cloud Storage (GCS) is Google Cloud Platform's object-storage service — GCP's equivalent of Amazon S3. It stores objects (up to 5 TB each) in buckets scoped to GCP projects and locations (regional, multi-region, dual-region), exposes a REST API + gRPC API, and provides eleven-9s durability via cross-region replication.

Why it shows up on this wiki

GCS appears in three distinct roles on this wiki:

• PlanetScale backup destination. Per the 2024-07-30 post (Source: sources/2026-04-21-planetscale-faster-backups-with-sharding): "It will store the backups themselves in either Amazon S3 or Google GCS, depending on which cloud the database cluster resides in."
• Iceberg-metadata + Parquet host for Redpanda Iceberg Topics on BYOC. Per the 2025-05-13 BYOC setup walkthrough (Source: sources/2025-05-13-redpanda-getting-started-with-iceberg-topics-on-redpanda-byoc): the Redpanda BYOC broker writes Iceberg table data (Parquet files + vN.metadata.json snapshot pointers) directly into a customer-owned GCS bucket, which BigQuery then reads via CREATE EXTERNAL TABLE — the canonical file-based catalog shape on GCS. The customer-bucket-ownership property is load-bearing for BYOC data ownership for Iceberg.
• Redpanda tiered-storage substrate. Per the 2025-06-20 GCP-outage retrospective (Source: sources/2025-06-20-redpanda-behind-the-scenes-redpanda-clouds-response-to-the-gcp-outage): GCS is the object-storage tier that receives asynchronous flushes of older Redpanda topic segments. The 2025-06-12 GCP outage produced elevated GCS PUT error rates that were customer-invisible because GCS is fallback, not primary in Redpanda's architecture — primary writes hit local NVMe + intra-cluster replication.

Seen in

• sources/2025-06-20-redpanda-behind-the-scenes-redpanda-clouds-response-to-the-gcp-outage — GCS as the tiered-storage substrate during Redpanda's 2025-06-12 incident retrospective. During the GCP global outage Redpanda observed "an increase in tiered storage errors... PUT requests were dominant" on GCS. No write-path impact because GCS is fallback, not primary in Redpanda's architecture — primary writes hit local NVMe + intra-cluster replication. Validates the async- flush-to-object-storage design by keeping elevated GCS error rates from cascading into customer write availability.

• sources/2025-05-13-redpanda-getting-started-with-iceberg-topics-on-redpanda-byoc — GCS as the customer-owned object-storage substrate for Redpanda BYOC Iceberg Topics. Broker writes Parquet + Iceberg metadata into the bucket; BigQuery reads via file-based catalog external-table DDL. No Redpanda-operated catalog in the query path.

• sources/2026-04-21-planetscale-faster-backups-with-sharding — canonical wiki disclosure of GCS as one of the two PlanetScale backup destinations (paired with S3 for AWS deployments). Stored backups are encrypted at rest.

Related

• systems/aws-s3
• systems/vtbackup
• systems/planetscale
• systems/redpanda — 2025-06-12 GCP outage retrospective canonicalises GCS as the tiered-storage substrate that elevated PUT error rates without impacting write availability.
• systems/redpanda-iceberg-topics · systems/redpanda-byoc — canonical wiki producer writing Iceberg into GCS.
• systems/google-bigquery — canonical wiki reader consuming Iceberg from GCS.
• systems/apache-iceberg — the table format hosted.
• concepts/iceberg-file-based-catalog · concepts/byoc-data-ownership-for-iceberg — the catalog shape and data-ownership compound property GCS participates in.
• concepts/tiered-storage-as-primary-fallback — the architectural property that made GCS error spikes customer-invisible during the 2025-06-12 outage.