With its innovative approach and groundbreaking compute density, TundraTM ES is radically improving how HPC and other large scale systems are being built and deployed. But compute is only part of the overall picture. Building highly effective systems requires storage that’s equally as innovative. With its family of unique building blocks, Tundra storage does just that, delivering flexible storage that can address balance, flexibility, usability, scalability and performance.
Balance – System architects need no longer be constrained by the limits of traditional monolithic storage solutions that require oversized infrastructure. With Tundra’s unique capability to locate storage and compute locally, we can design, build and deliver the right kind of storage for the problem in hand, optimized for performance and reliability, exactly where it can best be used.
Flexibility – Tundra storage building blocks can leverage drives from 3.5″ nearline to SSD/NVMe in units as small as a single 1 OU node, increasing in capability through 2 OU nodes, 2 OU JBOD/RAID shelves through to large capacity Fabric Attached Storage.
Usability – In conjunction with well known filesystem solutions such as Lustre, CEPH, NFS, GusterFS and others, we can design systems that deliver a broad range of functionality, not just for traditional HPC, but also for end-user applications such as Hadoop, OpenStack/Swift, etc.
Performance – Combined with Tundra’s Fabric Infrastructure, one can simply design storage solutions, for example HPC Lustre filesystems, that deliver as little as 1 GB/s to more than 100GB/s of aggregate filesystem bandwidth, in systems whose capacity can be designed to easily exceed 100’s of PetaBytes. In addition, Tundra’s ability to co-locate compute and storage, allows for the construction of large scale Big Data style systems.