As enterprises aggressively ramp up their AI initiatives in response to new business imperatives, the performance of their storage systems becomes increasingly critical, especially for their globally distributed file and object data. To address this challenge, our product and engineering teams have collaborated to deliver enhancements to our flagship object storage software and hardware solutions.
Today IBM is announcing IBM Storage Scale 5.2.1, which adds high-performance S3 object access to files and directories via Cluster Export Services for S3 (CES S3). It’s designed for data-intensive workflows that need S3 object access to data stored in Storage Scale filesystems, such as AI, analytics, and watsonx workloads.
CES S3 helps globally dispersed teams collaborate on data regardless of protocol, location, or format. It supports standard AWS S3 API calls and is optimized for multi-protocol data access (including GPUdirect support), to enable workflows which access the same instance of data using S3 object and other access protocols. Scale 5.2.1 delivers impressive performance – it’s been tested in systems up to 10 CES S3 nodes, up to 5,000 S3 accounts and 5,000 S3 buckets, and with up to 100 million objects supported per bucket.
Scale 5.2.1 provides support for the NVIDIA Bluefield 3 data processing unit (DPU), for emerging systems with the ability to offload and accelerate software-defined storage, networking, and other functions. The new release includes improved NFS filesystem resilience and scalability and better operator visibility into the causes of common network issues. It also includes user experience improvements to streamline the process of deploying and updating the software.
The software enhancements announced today will benefit customers running the IBM Storage Scale System 6000 hardware platform as well as those deploying Storage Scale on x86, IBM Power, IBM zSystem mainframes, ARM-based POSIX client, virtual machines, or Kubernetes.
Storage Scale 5.2.1 is available as of today.
#Featured-area-2