December 3rd, 2013

Scheduling with Uncertainty in Cloud and Grid Environment

Andrei Tchernykh

Parallel Computing Laboratory Lead CICESE Research Center


In this talk, we will discuss a model for cloud computing applications, called CA-DAG. This communication-aware model allows making separate resource allocation decisions, assigning processors to handle computing jobs, and network resources for data transmissions. We will discuss the benefits, weaknesses, and performance characteristics of such a model and resource allocation strategies in presence of uncertainty due to dynamic behavior of the execution context, job mix workloads, or uncertainty of the workflow properties.

In real dynamic environment, the execution environment of applications, and scheduling parameters can be changed over the time, we might not know, and cannot predict ahead: quantity of data that can be managed; quantity of computation required by a task; exact knowledge about the system; effective processor speed; number of available machines; their location; the network topology; effective communication bandwidth; etc. In general, an execution environment will differ for each program/service invocation. To understand how well the available scheduling solutions are applicable in the cloud computing environment, the main properties and requirements of cloud computing applications are reviewed.

Another drawback of existing scheduling strategies is a consideration of a single objective criterion. Most of them adopt the best-effort approach, which optimizes only performance objective. However, scheduling in the cloud is inherently multi-objective. For instance, system performance related objectives, cost based, energy consumption based, and QoS based objectives must be considered.

We also address scheduling algorithms for different scenarios of HPC, Grid and Cloud Infrastructures. We provide some theoretical and experimental bounds and QoS. Dynamic and adaptive approaches are presented.

About Andrei Tchernykh

Andrei Tchernykh is a researcher in the Computer Science Department, CICESE Research Center, Ensenada, Baja California, Mexico. From 1975 to 1990 he was with the Institute of Precise Mechanics and Computer Technology of the Russian Academy of Sciences (ITM&VT, Moscow, Russia). He received the Ph.D. in Computer Science from ITM&VT in 1986. In CICESE, he is a coordinator of the Parallel Computing Laboratory. He is member of the National System of Researchers of Mexico (SNI), Level II. He leads number of national and international research projects. He served as a program committee member of professional international conferences, and as a general chair for International Conferences on Parallel Computing Systems. His main interests include scheduling, load balancing, adaptive resource allocation, scalable energy-aware algorithms, green grid and cloud computing, eco-friendly P2P scheduling, multi-objective optimization, scheduling in real time systems, computational intelligence, heuristics, meta-heuristics, and incomplete information processing.