Google Cloud Storage

Use it when

• The platform runs on Google Cloud; BigQuery, Dataproc, and Dataflow read it natively and in-region traffic to Google services is free.
• Access patterns drift: Autoclass moves objects between Standard, Nearline, Coldline, and Archive automatically, removing the stale-lifecycle-policy failure mode.
• AI and ML I/O is intense enough for the Rapid class, or folder-aware buckets and FUSE mounts fit the tooling.
• S3-written tools must mostly work; the XML interoperability endpoint covers the common path with a credential change.

Think twice when

• Data leaves Google Cloud often; egress to the internet and other regions is charged, the same shape as S3 and Azure.
• Full S3 fidelity is assumed; the interoperability endpoint is partial, and edge cases surface with tools that lean on less common S3 behaviors.
• Cold data churns: Nearline, Coldline, and Archive carry 30-, 90-, and 365-day minimum durations.

How it runs

Fully managed with eleven-nines durability, erasure-coded across devices, in regional, dual-region, or multi-region buckets (12-hour RPO by default, 15 minutes with turbo replication). Retention locks, bucket lock, holds, and soft delete cover compliance. Events flow through Pub/Sub and Eventarc into Cloud Run. An always-free allowance covers 5 GB of Standard storage in three US regions.