The Dot Hill parallel distributed file system solution provides a massively scalable and high performance infrastructure for use in a broad range of high performance computing (HPC) industries such as meteorology, simulation, oil& gas, life science, rich media and finance. The Lustre file system is scalable to hundreds of petabytes (PB) of storage across multiple storage arrays and servers. More than a terabyte per second (TB/s) of aggregate I/Othroughput is possible with a properly sized configuration.
Aside from scaling to support hundreds of terabytes to petabytes of storage, the solution is unique in that it is built specifically to address the distinctive signature HPC workloads that consist of many nodes or threads performing sequential access of very large files; while the workload of theindividual node or thread is sequential in nature, the composite workload of many nodes or threads causes the underlying access pattern to become highlyrandomised.
“Extreme data analysis and data modeling is being employed by scientists and technologists who are tasked with solving the world’s toughest issues. Dot Hill storage has been engineered to maintain high performance for randomised sequential workloads. Data storage systems must adapt to this dynamic landscape by offering solutions that provide density, scalability and performance, and yet are simple to manage and affordable,” said Bill Wuertz, senior vice president of products and solutions, Dot Hill.
The backbone of the solution is the Dot Hill Ultra48™ AssuredSAN® 4004 and newly launched Ultra56™ AssuredSAN 4004 storage arrays, which are designed to support both high performance and reliability; two separate arrays function to preserve and manage access to both meta and primary data. The Ultra48 AssuredSAN 4004 provides high performance small form factor SSD and disk drives, which are ideal for metadata operations; and the Ultra56 AssuredSAN4004 supports up to 1.3PB with both flash and high capacity large form factor drives, which are ideal for Lustre’s OST data. Adaptive caching technologies are employed to accommodate dozens and even hundreds of independent streams of data without degrading overall throughput – critical for reliable, high performance data streaming. In addition, the proven 99.999 percent availability of AssuredSAN products virtually eliminates downtime.
The Intel® Enterprise Edition for Lustre software provides the high performance distributed file system and namespace component of the solution, which supports bandwidth, performance and scaling above and beyond traditional storage technology. The Intel® Enterprise Edition for Lustre* software includes a rich set of management tools and a unified namespace, eliminating the need to micromanage storage pools.
“High performance computing is a powerful competitive advantage for today’s enterprises. Today, maximising investments in HPC is measured in application-level performance, which requires proven scale-out storage solutions to overcome the limited performance and limited productivity common to most storage solutions today. Purpose-built for very high speed at massive scale, Intel Enterprise Edition for Lustre software makes storage simpler to manage and more productive,” said Brent Gorda, general manager of Intel’s High Performance Data Division.
Mellanox InfiniBand high-speed networking components and infrastructure support enhanced data rates up to 100Gbps, which is the highest throughput and lowest latency interconnect available on the market today. In addition, Mellanox network solutions utilise remote direct memory access (RDMA) protocols to make ultra-efficient use of computing resources. Using RDMA over InfiniBand, data transfer latencies can be reduced by more than 90 percent and CPU efficiencies can by elevated up to 96 percent.
Key Benefits of the Dot Hill Scale-Out, High Performance Lustre File Serving Solution
The solution provides exceptional scalability, linear performance, high availability and data protection. Read more about the solution.
· Scalability is simplified through the addition of storage modules as needed to expand to petabytes of usable capacity. Each individual storage module can be configured with as much as 576TB of raw capacity
· Performance of up to 6GB/s throughput is achieved with each individual storage module, andscaled linearly with added modules. InfiniBand network paths deliver up to 56Gbps throughput to each compute node.
· High availability is supported through redundant fail-over components as well as the ability to distribute the file system across multiple servers.
· Protection of valuable data sets is achieved through the use of enterprise-class components and drives, along with RAID technology.