Server- and Core-Level Renewable-Energy Allocation Under Uneven Workload Utilization

Exploit resource-management opportunities to improve renewable-energy utilization under server-level or core-level energy allocations when dynamic cloud workloads are unevenly distributed across servers, particularly for applications with strict latency service-level objectives that prevent workload migration.

Background

Latency-sensitive cloud applications may run in distributed, resource-constrained environments such as colocation data centers, where tenants share limited power budgets and renewable-energy resources. Existing carbon-aware approaches commonly allocate renewable energy at the data-center level, enabling workload management across servers or clouds as renewable capacity changes over time and location.

For applications with strict latency SLOs, migrating workloads may be infeasible, motivating server-level or core-level renewable-energy allocation. However, dynamically varying workloads can be distributed unevenly across servers, causing renewable-energy capacity assigned to individual servers or cores to be underutilized. The unresolved problem is to devise resource-management techniques that address this underutilization without violating strict latency requirements.

References

Exploiting opportunities in resource management to improve such is an interesting open problem.

— Carbon-aware Resource Management for Latency-Sensitive Cloud Computing Environments: A Taxonomy and Future Directions  (2609.03270 - Hewage et al., 3 Sep 2026) in Section 5.1, Cloud Environment Characteristics

Leveraging such with application fault-tolerance for renewable energy intermittency is still an open challenge.

— Carbon-aware Resource Management for Latency-Sensitive Cloud Computing Environments: A Taxonomy and Future Directions  (2609.03270 - Hewage et al., 3 Sep 2026) in Section 5.2, Application Characteristics and Latency SLOs