Cloud Storage Rapid adds cache ingestion during writes

Google Cloud detailed a new ingest-on-write capability for Rapid Cache on May 12. In its Cloud Storage Rapid update, the company describes populating the cache while data is written to a Cloud Storage bucket, addressing workloads that read newly produced objects immediately.

A change to the first read

Previously, an initial read triggered cache ingestion and came from the bucket at standard performance, according to Google. Ingest on write changes when the cache receives the data. The update highlights checkpoint restoration and pipelines where one processing stage produces data for the next.

The Rapid product family had already been announced at Cloud Next 2026. This May article explains its capabilities and identifies ingest on write as a new feature; it is not evidence that the entire storage family launched on that date.

Test the producer-to-consumer interval

The useful measurement is the time between producing usable data and completing the next stage that needs it. A cache benchmark that repeatedly reads an already warm object can miss the exact problem this feature addresses.

An evaluation should therefore include fresh object writes followed by immediate reads, using the object sizes and concurrency expected in production. Compare complete pipeline duration, downstream waiting time, and the cost of keeping the relevant data close to compute.

Also decide which writes deserve proactive ingestion. Frequently written data that is rarely read presents a different tradeoff from a checkpoint needed promptly after an interruption. Retention and recovery requirements should be evaluated separately from access speed.

For teams investigating idle compute, this update suggests a specific experiment: measure the first read after a write. If that transition dominates waiting time, changing cache ingestion may be more relevant than optimizing an already fast steady-state read.