| The development of new scalable algorithms for multi-way partitioning, scheduling and resource management for Data-Centers (DC) dealing with large groups of applications distributed over large sets of heterogeneous resources. Algorithms are aimed on optimization of various goals including throughput, resource utilization, power consumption, etc.
Almost all of the developed systems were transfered to various product groups, incorporated into internal or external products, and deployed.
New optimization strategies have attained significant improvements in diverse DC objectives, including a two-times improvement in resource utilization, 30%-35% decrease in power consumption, etc. |
