Title:Optimizing egalitarian performance in the side-effects model of colocation for data~center resource management

Abstract:In data centers, up to dozens of tasks are colocated on a single physical machine. Machines are used more efficiently, but tasks' performance deteriorates, as colocated tasks compete for shared resources. As tasks are heterogeneous (CPU-, memory-, network- or disk-intensive), the resulting performance dependencies are complex.
We explore a new combinatorial optimization model that uses two parameters of a task - its size and its type - to characterize how a task influences the performance of the other tasks allocated to the same machine. We study the egalitarian optimization goal: maximizing the worst-off performance. This problem generalizes the classic makespan minimization on multiple processors (P||Cmax).
We prove that polynomially-solvable variants of multiprocessor scheduling become NP-hard and hard to approximate when the number of types is not constant. We propose a PTAS and a series of fast approximation algorithms when the number of types is constant. By simulation on instances derived from a trace of one of Google clusters, we show that our algorithms that take into account types lead to lower costs compared with P||Cmax baseline.
The notion of type enables us to model degeneration of performance caused by colocation using standard combinatorial optimization methods. Types add a layer of additional complexity. However, our results - approximation algorithms and good average-case performance - show that types can be handled efficiently.