SAS ® Grid Manager I/O: Optimizing SAS ® Application Data Availability for the Grid
Gregg Rohaly, Harry A. Seifert · 2014
As organizations deploy SAS ® applications to produce the analytical results that are critical for solid decision making, they are turning to distributed grid computing operated by SAS ® Grid Manager. SAS Grid Manager provides a flexible, centrally managed computing environment for processing large volumes of data for analytical applications. Exceptional storage performance is one of the most critical components of implementing SAS in a distributed grid environment. When the storage subsystem is not designed properly or implemented correctly, SAS applications do not perform well, thereby reducing a key advantage of moving to grid computing. Therefore, a well-architected SAS environment with a high-performance storage environment is integral to clients getting the most out of their investment. This paper introduces concepts from software storage virtualization in the cloud for the generalized SAS Grid Manager architecture, highlights platform and enterprise architecture considerations, and uses the most popularly selected distributed file system, IBM GPFS, as an example. File system scalability considerations, configuration details, and tuning suggestions are provided in a manner that can be applied to a client’s own environment. A summary checklist of important factors to consider when architecting and deploying a shared, distributed file system is provided.