A time-sliced, cyclically staggered scheduling strategy, partial disaggregation, is shown to improve SLO-aware goodput for 30B to 72B LLMs on commodity-interconnect clusters relative to co-located and fully disaggregated baselines.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.DC 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
EcoServe: Enabling Cost-effective LLM Serving with Proactive Intra- and Inter-Instance Orchestration
A time-sliced, cyclically staggered scheduling strategy, partial disaggregation, is shown to improve SLO-aware goodput for 30B to 72B LLMs on commodity-interconnect clusters relative to co-located and fully disaggregated baselines.