Optimization-based resource allocation in large-scale systems often must trade-off responsiveness and allocation quality. Generally, allocations are reconsidered every few minutes (a round) by formulating and solving a new optimization problem. This paper introduces continual optimization, which reframes round-based resource allocation as a sequence of interconnected problems, leveraging the observation that these resource allocation problems often only change by small amounts across successive rounds to reduce solving times. COpter provides a method for continual optimization of Linear Programs (LP) and Mixed Integer Linear Programs (MILP) formulations of resource allocation problems by combining three innovations: (1) an efficient-to-update problem representation for incremental changes, (2) a proximal-point method implementation that can provably benefit from prior computational effort and allocations, and (3) lightweight heuristics for mixed-integer problems that recover feasible integer solutions with negligible quality loss. We evaluate COpter on problems in three domains: GPU cluster scheduling, shard load balancing, and WAN traffic engineering. Overall, we find that COpter finds high-quality solutions while reducing solver runtimes by 57–83× compared to state-of-the-art commercial solvers. Compared to problem partitioning approaches (POP), COpter simultaneously improves allocation quality and reduces end-to-end allocator runtimes by 1.5–30×.

The Sia scheduler efficiently assigns heterogeneous deep learning (DL) cluster resources to elastic resource-adaptive jobs. Although some recent schedulers address one aspect or another (e.g., heterogeneity or resource-adaptivity), none addresses all and most scale poorly to large clusters and/or heavy workloads even without the full complexity of the combined scheduling problem. Sia introduces a new scheduling formulation that can scale to the search-space sizes and intentionally match jobs and their configurations to GPU types and counts, while adapting to changes in cluster load and job mix over time. Sia also introduces a low-profiling-overhead approach to bootstrapping (for each new job) throughput models used to evaluate possible resource assignments, and it is the first cluster scheduler to support elastic scaling of hybrid parallel jobs.