MWS Cloud has introduced a specialized "Warm" object storage class, optimized for Big Data workloads in artificial intelligence and machine learning projects. The solution is based on high-speed NVMe drives and provides data transfer speeds of up to 1.8 GiB/s with a first-byte latency of approximately 20 ms.
What Happened
MWS Cloud has launched a new S3-compatible "Warm" storage class specifically designed for AI, ML, and Big Data tasks. The system is built on NVMe drives, supports encryption, replication between two Tier III data centers, and access management via IAM. The storage is fully compatible with the AWS S3 API, allowing the use of familiar tools such as AWS CLI or boto3.
Context
Effective neural network training and large-scale data processing require ultra-low read latencies (Data Loading). Using NVMe drives in cloud S3 storage significantly reduces data waiting times compared to standard solutions based on HDD or conventional SSDs.
Why It Matters for the Industry
The emergence of specialized high-speed S3-compatible NVMe storage within the Russian Federation reduces the need for expensive local infrastructure. This allows companies to deploy ML experiments faster, minimizing latency when working with heavy datasets and creating a foundation for establishing AI-ready cloud storage standards in the region.
Why It Matters for Users
Developers and ML startups gain the ability to work with massive datasets without the need to purchase and maintain their own high-speed NVMe hardware. The pricing model allows for scaling computing power and experiments in the cloud using existing software tools for S3.
What Is Not Yet Known / Limitations
The official announcement lacks detailed testing methodologies and benchmarks, which requires cautious interpretation of the claimed speeds of up to 1.8 GiB/s until they are fully verified.
Sources
Author
Look at AI, Editorial Team