Optimizing Cloud Computing Performance

The field of cloud computing is moving towards optimizing performance in large-scale clusters and data-center networks. Researchers are focusing on devising effective dispatching policies, load balancing strategies, and routing algorithms to minimize congestion and reduce response times. Innovative approaches, such as two-stage architectures and hierarchical forecast reconciliation, are being explored to improve the efficiency of cloud systems. Noteworthy papers include:

  • Two-Stagification, which introduces a two-stage cluster architecture to improve mean response times.
  • Hierarchical Forecast Reconciliation on Networks, which reformulates hierarchical forecast reconciliation as a network flow optimization to enable forecasting on generalized network structures.

Sources

"Two-Stagification": Job Dispatching in Large-Scale Clusters via a Two-Stage Architecture

TailBench++: Flexible Multi-Client, Multi-Server Benchmarking for Latency-Critical Workloads

Dynamic load balancing for cloud systems under heterogeneous setup delays

Minimum Congestion Routing of Unsplittable Flows in Data-Center Networks

Hierarchical Forecast Reconciliation on Networks: A Network Flow Optimization Formulation

Multiserver-job Response Time under Multilevel Scaling

Built with on top of