Introducing Cloud Storage Rapid
At Google Cloud Next '26, an important development was unveiled: the Cloud Storage Rapid suite. This new offering aims to cater specifically to data-heavy tasks like AI and analytics. It combines high-performance object storage capabilities with a focus on meeting the demands of modern workloads, which has grown exponentially in recent years.
Cloud Storage Rapid encompasses two key components: the Rapid Bucket and Rapid Cache, both engineered to resolve the growing performance limitations currently facing AI and machine learning (ML) practitioners.
Understanding Rapid Storage Needs
The rise of AI has fundamentally changed how organizations process and analyze data. As teams tackle multi-trillion parameter models and deploy inference across global networks, the importance of effective data storage becomes underscored. While much attention is placed on GPU and TPU accelerators, the unsung hero that often hinders performance is storage itself.
Data storage serves as the backbone for these accelerators during both training and real-time inference. When clustered AI/ML systems stall waiting for data reads or are delayed by checkpoint writes, organizations find themselves wasting computational resources. In light of these challenges, many developers face a difficult choice: they can either opt for niche zonal storage with specialized performance or rely on the global capabilities of a service like Google Cloud Storage, which may lack the required speed for cutting-edge workloads.
With Cloud Storage Rapid, the goal is to eliminate these bottlenecks by offering alternatives that co-locate compute tasks with high-speed zonal storage. This synergy ensures that GPUs and TPUs operate at full capacity without encountering unnecessary delays.
Features of Cloud Storage Rapid
**Rapid Bucket**
The Rapid Bucket represents a groundbreaking approach, particularly suited for large-scale generative AI projects and complex analytics tasks. It harnesses the Colossus distributed storage framework—an integral part of services like Gemini and YouTube—to facilitate unprecedented read and write performance in a dedicated zonal repository.
Users can expect ultra-low latency and scalability that outpaces traditional storage solutions. For instance, Rapid Bucket can handle up to 20 million queries per second while achieving sub-millisecond latency. Highlights include:
- **Extreme Scalability**: Up to 15+ TB/s aggregate read throughput from a single bucket.
- **Enhanced Performance**: Introduction of native appends and vectored reads makes for a more efficient system.
**Rapid Cache**
Rapid Cache is designed to streamline operations for existing Cloud Storage buckets, boosting read speeds without requiring modifications to applications. Initially unveiled at Cloud Next '25, it has quickly gained traction within the AI/ML community. With a staggering read throughput of 2.5 TB/s, it can enhance tasks such as data preparation and model inference loading by over 114%.
One of the standout capabilities of Rapid Cache is the "ingest on write" feature. This allows data to be rapidly ingested into the cache at the same moment it’s written to the Cloud Storage bucket, thereby reducing time lost to caching delays and yielding faster recovery times for workloads affected by interruptions.
Real-World Applications
For organizations like Thinking Machines Lab, the adoption of Rapid Cache has transformed their data infrastructure. James Sun, a technical staff member at the lab, highlighted their integration of Rapid Cache throughout their AI/ML processes. The adaptability of this solution allows Thinking Machines to efficiently manage their diverse workloads—whether for data processing, training, or activating models via Tinker, their flexible API for model fine-tuning.
As they faced challenges like managing complex data architectures and the potential for operational bottlenecks, Rapid Cache emerged as a vital tool. Sun reported that the technology has significantly decreased error rates and improved overall stability, enabling smoother multi-modal training processes.
Conclusion
Cloud Storage Rapid provides a compelling advancement for anyone involved in high-performance data applications. Both Rapid Bucket and Rapid Cache strike a balance between speed, efficiency, and reliability, addressing contemporary challenges in data management head on.
If you’re responsible for deploying data-intensive workloads or AI applications, looking into the Cloud Storage Rapid offerings might be your next strategic move. The integration promises operational efficiency while ensuring your capabilities remain on the cutting edge. Explore the
Cloud Storage Rapid family today.