• FLASH CACHING
      TECHNOLOGY
      DESIGNED FOR BIG DATA

Storage Fusion Xcelerator (SFX)

READ CACHE ACCELERATOR FOR PARALLEL FILE SYSTEMS

SFX leverages hybrid media and application aware technology to intelligently accelerate parallel file systems. Our read cache accelerator consists of inexpensive, commodity SSDs placed directly in your SFA Storage System that accelerate I/O performance for both GPFS™ & Lustre® by reducing the time CPUs are idle waiting for data from spinning disk.
SFX is both intelligent and highly efficient. This application-aware software dynamically provides out-of-band hints that can allocate more control to your applications. This enables the file system to proactively pre-stage the blocks from spinning disk that will benefit from the cache accelerator and significantly reduce latency.
HIGH PERFORMANCE
Reduce IO bottlenecks and perform
at speeds much closer to RAM
with up to 300x more IOPS.
LOWER COST
The right combination of SFX, SSDs
and HDDs delivers the performance
promise of flash at HDD prices.
BETTER TCO
Optimize solution performance
and capacity to ensure lowest
total cost of ownership.

PROBLEM: THE SERVER-STORAGE
PRICE AND PERFORMANCE GAP

In today’s market, flash memory commands premium pricing as it offers orders of magnitude improvements in reducing latency and increasing IOPS. At the same time, HDDs offer a lower price point, but at lower IOPS than flash. Because of this performance gap, the result is either overspending on high performance flash or deploying a lower performance system to meet budget requirements.
Cache Accelerator

SOLUTION: SFX READ CACHE ACCELERATOR FOR PARALLEL FILE SYSTEMS

SFX leverages hybrid media and application aware technology to intelligently accelerate parallel file systems. Our read cache accelerator consists of inexpensive, commodity SSDs placed directly in your SFA Storage System that accelerate I/O performance for both GPFS™ & Lustre® by reducing the time CPUs are idle waiting for data from spinning disk.

SFX is both intelligent and highly efficient. This application-aware software dynamically provides out-of-band hints that can allocate more control to your applications. This enables the file system to proactively pre-stage the blocks from spinning disk that will benefit from the cache accelerator and significantly reduce latency.

SFX READ MODE

Designed for read intensive workloads. SFX caches frequently accessed data sets in the fastest tier to significantly speed up application performance. (ex. Genomics, Oil & Gas)

SFX INSTANT COMMIT MODE

Designed for read after writes patterns. It populates the SFX tier with new writes to “warm-up” cache as well as accelerating subsequent reads. (ex. CDN)

SFX CONTEXT COMMIT MODE

Allows applications and file systems to send down “hints” to the storage system programmatically or via command line. This delivers the best possible storage acceleration by eliminating the need to “guess” what the IO pattern will be upon deployment. (ex. Web, Databases, Parallel File Systems)

ACCELERATING PARALLEL FILE SYSTEMS


GRIDScaler
Enterprise Parallel File System

EXAScaler
Parallel File System
Parallel file systems are a natural fit for streaming and large file access, but are not well known for small file performance. DDN has solved this challenge to deliver the industry’s most advanced GPFS & Lustre parallel file system appliances. Powered by SFX Read Acceleration, both GRIDScaler & EXAScaler enjoy dramatically improved metadata performance and small file read I/O rates. Applications direct GRIDScaler or EXAScaler to load frequently accessed data into SFX cache, ensuring the most critical data is readily available for the most demanding applications.