Home > Storage > PowerScale (Isilon) > Industry Solutions and Verticals > Analytics > Deep Learning with Dell EMC Isilon > File reads
For files marked with an access pattern of concurrent or streaming, OneFS can take advantage of pre-fetching of data based on heuristics used by the Isilon SmartRead component. SmartRead can create a data pipeline from L2 cache, prefetching into a local L1 cache on the captain node. This greatly improves sequential-read performance across all protocols and means that reads come directly from RAM within milliseconds. For high-sequential cases, SmartRead can very aggressively prefetch ahead, allowing reads or writes of individual files at very high data rates.
SmartRead intelligent caching allows for very high read performance with high levels of concurrent access. Importantly, it is faster for Node 1 to get file data from the cache of Node 2 (over the low-latency cluster interconnect) than to access its own local disk. SmartRead algorithms control how aggressive the pre-fetching is (disabling pre-fetch for random-access cases) and how long data stays in the cache, and optimizes where data is cached.