Abstract
The isolated optimization of distinct data center fleets may lead to systemic inefficiencies and may not optimally utilize the collective compute capacity available across different infrastructures, creating challenges in cost management and reliability as well as water and energy consumption. Systems and methods are described that can provide a framework for a federated marketplace where compute demand can be expressed through standardized bids and available capacity can be advertised through standardized offers. A central optimization engine can ingest these bids and offers to match workloads with suitable data center resources based on a plurality of criteria, such as cost, latency, hardware requirements, and sustainability metrics. This co-optimization across multiple, independently-owned infrastructures may facilitate the creation of a more fungible pool of compute resources, which can allow workloads to be dynamically placed to improve cost-effectiveness and overall resource utilization.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Smith, Raiford; Radovanovic, Ana; Vikram, Anant; and Malina, Roman, "Co-Optimization of Compute Workloads Across Heterogeneous Data Center Fleets", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/9508