Home > Storage > PowerScale (Isilon) > Product Documentation > Storage (general) > PowerScale OneFS SmartFlash > OneFS caching
The caching infrastructure enables OneFS to deliver extreme levels of performance, while maintaining globally coherent read and write access across the Dell PowerScale cluster. OneFS goes beyond traditional storage caching architectures in that it was designed to take advantage of the system’s distributed and highly parallel nature. Traditional scale-up storage systems typically have a fixed amount of cache per “head” unit (filer or server). The OneFS powered clustered architecture ensures that as storage grows, so does the amount of cache available to the system. Furthermore, unlike traditional NAS and SAN systems, OneFS caching is globally coherent across the cluster. Cached data is available to any node in the cluster, regardless of which node it physically resides on. Cache sharing across a consolidated storage environment is critical for today’s large-scale storage deployments, in terms of performance, scalability, efficiency, and overall cost of ownership.
Cache is typically used in storage systems to reduce the latency associated with retrieving data from disk, improving performance for clients making read requests. It is also used to improve the performance of writes, by acknowledging the write to the client when it has been written to cache. It later writes the data in the cache to disk.
The OneFS caching infrastructure leverages the system memory (RAM) and nonvolatile memory (NVRAM) in each node. It uses solid state drives (SSDs) and the increased capacity, affordability, and storage persistence that they offer.
Caching occurs in OneFS at multiple levels and for various types of data. For this discussion, we focus on caching of file system structures in main memory and on SSD.