REMOP: REmote-Memory-aware OPerator Optimization

2026-06-17

AI Summary

This work addresses the inefficiency of traditional database operators in remote memory environments, where ignoring communication round-trip overhead leads to poor performance during spilling. The paper presents the first operator optimization framework tailored for remote memory, incorporating the number of communication rounds into an operator-level latency cost model—thereby moving beyond the conventional I/O-centric optimization paradigm. It introduces round-trip-aware memory buffer partitioning strategies for key operators such as block nested-loop join, external merge sort, and external hash join. Implemented in DuckDB, the approach reduces communication rounds by up to 97% and operator execution time by up to 48% on a two-node platform. End-to-end evaluation on TPC-H and TPC-DS spilling queries demonstrates average speedups of 22.7% and 26.4%, respectively.

Abstract

Remote and disaggregated memory tiers expand the effective memory capacity of analytical database engines, but they also reshape the cost structure of out-of-memory query processing. When an operator spills beyond local DRAM, moving pages to remote memory incurs both data-transfer time and a fixed round-trip latency per transfer. Classical operator analyses and buffer-allocation heuristics primarily target disk spilling by minimizing total I/O volume. Under remote memory, these strategies can be suboptimal because they may trigger excessive transfer rounds. We present REMOP, a remote-memory-aware operator optimization framework that uses transfer-round-aware intra-operator memory policies to improve out-of-memory execution under tight memory budgets. REMOP introduces the number of transfer rounds into the latency cost model and derives operator-specific buffer-partitioning strategies, instantiating the approach for blocked nested-loop join, external merge sort, and external hash join in DuckDB. Our evaluation on a two-node compute-memory testbed shows that REMOP reduces transfer rounds by up to 97% and operator runtime by up to 48% on spill-heavy microbenchmarks, and lowers the average runtime of spilling TPC-H and TPC-DS queries by 22.7% and 26.4% end-to-end.

Problem

• remote memory
• operator optimization
• transfer rounds
• out-of-memory processing
• buffer allocation

Innovation

• remote memory
• transfer rounds
• operator optimization
• memory-aware scheduling
• out-of-memory processing