LaissezCloud: Continuous Resource Renegotiation for the Public Cloud

Public clouds increasingly rely on heterogeneous hardware, but their allocation models—on-demand and spot instances—remain rigid and inefficient for dynamic workloads. LaissezCloud, a new platform from researchers Tejas Harith and Antoine Kaufmann, introduces continuous resource renegotiation during execution. Unlike spot instances that use launch-time bids and unilateral preemption, LaissezCloud keeps allocations contestable: tenants and operators update bids online, and a tenant retains a resource only as long as its bid exceeds competing demand. This creates a fluid market where pricing serves as both a narrow waist and an incentive-alignment mechanism between untrusted participants. Tenants express utility through bids, while operators price in power, cooling, or carbon constraints without exposing internal telemetry.

Across a diverse set of accelerator workloads, LaissezCloud reduces performance degradation under contention by 8-23% compared to on-demand and spot baselines. The system scales to clusters of at least 10,000 nodes, making it viable for large-scale deployments. By decoupling allocation from rigid service classes, LaissezCloud enables more efficient resource use in oversubscribed, heterogeneous clusters, addressing a key pain point for cloud operators and tenants alike. The paper is available on arXiv and is pending registration with DataCite.