Home > Workload Solutions > High Performance Computing > White Papers > Dell Validated Design for HPC pixstor Storage—Joint Solution with Kalray > Solution introduction
In high-performance computing (HPC), all its different components must be balanced for optimal performance and to avoid bottlenecks. Technological evolution of compute node components like higher core-count CPUs, larger and faster memory, faster PCIe buses, and increasingly faster networks caused storage to become the bottleneck in many workloads. Parallel File Systems (PFS) can scale to meet greater performance demands, providing concurrent access to a single file or a set of files from multiple nodes, efficiently and securely distributing data to multiple LUNs across several storage servers and storage arrays. These PFS normally use spinning media to provide the highest capacity at the lowest cost. However, often the speed and latency of spinning media cannot keep up with the demands of many modern HPC workloads. Recent rapid advances in storage technologies like Non-Volatile Memory Express (NVMe) SSDs, make available better and faster options to PFS storage systems. The use of this NVMe technology in the form of burst buffers, faster tiers, or even fast scratch (local or distributed) can mitigate or eliminate spinning media issues.
A well-balanced storage system is required to achieve optimal performance. Its components such as back-end arrays, storage controllers, disk drives, IO cards, network adapters, and switches must be able to provide similar bandwidth and distribute the load among the processors and memory of the servers they use. The PFS and supporting software must be able to balance the load on the different storage system components distributing data among them, monitoring performance along with the health of the components, and provide administration tools to manage the solution efficiently.
The Dell Validated Design for HPC pixstor Storage is a highly available (HA) solution. It offers a cost-effective, high-capacity tier-based on Dell PowerEdge R650, R750, and R7525 servers and PowerVault ME5084 storage arrays arranged in ME5 storage modules, and NVMe nodes to address high-bandwidth demands. Each ME5 storage module can scale up its capacity to 11.93 PiB of formatted and available storage (raw capacity 13,440 TB) and can scale out to customer requirements by adding more modules up to the required capacity and performance. For situations that require high throughput or metadata demands, an NVMe tier can scale out performance by adding pairs of HA nodes with NVMe devices using replication among the server pairs.
pixstor storage uses the well-established General Parallel File System (GPFS), now known as Spectrum Scale, with a full range of enterprise features that can be geo-distributed, components that simplify administration, additional advanced search capabilities, extensive connectivity using enterprise protocols, and access to other storage devices both on and off premises using cloud protocols and other enterprise protocols. This storage solution is fully supported (hardware and software) with deployment services available and offers single namespace, multi-tiered, easy-to-use, high throughput, among other features.
This document describes the architecture, general design guidelines, and performance of the HPC pixstor storage solution for large transfers Sequential, small transfers Random IO, and Metadata operations workloads.
Note: The contents of this document are valid for the described software and hardware versions. For information about updated configurations for newer software and hardware versions, contact your Dell Technologies sales representative.
This document is primarily intended for solution architects and administrators who plan to use or are using HPC clusters and require high-performance storage that keeps up with modern processing power. This document is not a detailed configuration guide but provides configuration best practices. It assumes that the readers understand the basic concepts of storage performance benchmarking.