Cited 0 time in webofscience Cited 4 time in scopus

Reducing Tail Latency of DNN-based Recommender Systems using In-storage Processing

Title
Reducing Tail Latency of DNN-based Recommender Systems using In-storage Processing
Authors
Kim, MinsubLee, Sungjin
DGIST Authors
Lee, Sungjin
Issue Date
2020-08-25
Citation
11th ACM SIGOPS Asia-Pacific Workshop on Systems, APSys 2020, 90-97
Type
Conference
ISBN
9781450380690
Abstract
Most recommender systems are designed to comply with service level agreement (SLA) because prompt response to users' requests is the most important factor that decides the quality of service. Existing recommender systems, however, seriously suffer from long tail latency when the embedding tables cannot be entirely loaded in the main memory. In this paper, we propose a new SSD architecture called EMB-SSD, which mitigates the tail latency problem of recommender systems by leveraging in-storage processing. By offloading the data-intensive parts of the recommendation algorithm into an SSD, EMB-SSD not only reduces the data traffic between the host and the SSD, but also lowers software overheads caused by deep I/O stacks. Results show that EMB-SSD exhibits 47% and 25% shorter 99th percentile latency and average latency, respectively, over existing systems. © 2020 ACM.
URI
http://hdl.handle.net/20.500.11750/12876
DOI
10.1145/3409963.3410501
Publisher
Association for Computing Machinery
Related Researcher
  • Author Lee, Sungjin Data-Intensive Computing Systems Laboratory
  • Research Interests Computer System, System Software, Storage System, Non-volatile Memory, Flash-based SSD, Distributed Storage Systems
Files:
There are no files associated with this item.
Collection:
Department of Electrical Engineering and Computer ScienceData-Intensive Computing Systems Laboratory2. Conference Papers


qrcode mendeley

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

BROWSE