Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
September 5, 2025Communications of the ACMOpen Access

TMO: Transparent Memory Offloading in Datacenters

View Full Paper
Ask AI
Bookmark
Share

Authors

JWJohannes WeinerNANiket AgarwalDSDan Schatzberg

Discussion

Loading...

Member takes

Overview

TMO finds optimal memory offloading strategies for datacenters, reducing costs and performance variance.

Key Points

  • TMO reduces total memory usage by 20–32% in datacenters, significantly lowering costs and improving efficiency.
  • The innovative Linux kernel mechanism identifies real-time performance stalls, guiding effective memory offloading decisions.
  • This solution leverages heterogeneous memory technologies, optimizing the offloading of both anonymous memory and file cache.
  • TMO has been in successful production since 2021, demonstrating its effectiveness across a vast hyperscale datacenter fleet.

Cite This Study

Weiner et al. (2025) studied this question.

synapsesocial.com/papers/68bb3a352b87ece8dc954c75https://doi.org/10.1145/3746651
View Full Paper
Ask AI
Bookmark
Share