Parallel Garbage collection in Solid State Drives
Halls department, Hall 4
Wednesday, 28 December 2016
11:30 - 12:30
In the last decade, NAND Flash-based SSDs have been widely adopted for high-end enterprise systems in an attempt to provide a high-performance and reliable storage. However, inferior performance is frequently attained mainly due to needing for Garbage Collection (GC). GC in flash memory is the process of identifying and clearing the blocks of unneeded data to create space for the new data to be allocated. GC is high-latency operation and once it is scheduled it can increase latency for later arriving I/O requests. On the other hands, SSDs have high levels of parallelism provided by channels, chips, dies and planes, which is sometimes under-utilized due to the resource contention. In this work, we propose a novel GC strategy, which leverages plane-level parallelism to improve GC latency and reduce performance inconsistencies.
Narges Shahidi received her B.Sc and M.Sc from Computer Engineering department at Sharif University of Technology at 2008, and 2010 respectively. She is currently a PhD student at Pennsylvania State University. Her research is focused on NAND Flash Solid State Drives. During her PhD she interned at several academic and industry labs, including Camel Lab at University of Texas at Dallas, Memory Solution Lab at Samsung Semiconductor Inc. (SSI) and Google Inc.