.png){kind=icon}
.png){kind=icon}
Return
Lemonade: Learning-based Heterogeneous Metadata Offloading for Disaggregated Memory
.png){kind=icon}
OA
AI
DOI：10.1145/3761807.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Direct Access (DA) in Disaggregated Memory (DM) is a promising solution that meets the high-performance requirements of AI applications. However, it lacks effective support for metadata management, making meta-data operations the major bottleneck. To address this, we propose Lemonade, a learning-based heterogeneous metadata offloading for disaggregated memory. Lemonade splits the metadata into highly regular and irregular ones, thus offloading the former into the client to avoid remote queries and enabling request redirection in the SmartNIC for the latter to ensure cost-effective correction and updates. Evaluations under microbenchmark and YCSB workloads indicate that Lemonade reduces latency by 72.8% and achieves a 1.43x increase in throughput compared to the state-of-the-art systems.
Keywords:
Disaggregated memory
direct access
heterogeneous metadata offloading
learning-based compression
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
2.6
Papers:
227
Citations:
2.3K
Organization
No organization information available