Collaborative Research Centre 901 On-The-Fly Computing

Here you find the project page of the CRC 901.

The objective of the Collaborative Research Centre 901 On-The-Fly Computing (CRC 901: OTF Computing) is the development of techniques and methods for automatic on-the-fly configuration and execution of individual IT applications. These applications consist of services that are traded on global markets. We consider in particular:

• the configuration of applications by service providers
• the execution of these applications at data centers and
• methods for quality assurance and protection of market participants.

Our vision of On-The-Fly Computing are individually and automatically configured IT services consisting of freely traded and flexibly combined services from global markets. The aim is a transparent combination of hardware and software from different vendors with minimal human interaction.

Structurally, the CRC is organized in three project areas: Project Area A deals with algorithmic and economic foundations for the organization of large, dynamic markets. Project Area B explores methods for modeling, analysis, composition, and quality of services as well as service configurations. Section C develops reliable execution environments for On-The-Fly Computing, and deals with issues of robustness and security of markets, the organization of highly heterogeneous OTF Compute Centers, and the execution of the configured services by such centers.

Research areas of the working group

Our research group is working within the framework of the CRC on the algorithmic foundations of the sub-projects A1, A3, and C2:

Project A1: "Possibilities and limits of local strategies in dynamic networks"

We are engaged in Project A1 with local strategies to address fundamental challenges for the organization of large, dynamic markets. Our studies encompass the design, analysis, and evaluation of local algorithms. Important issues are the role assignment, the search for services, and the group management. We are also working on a complexity-theoretic classification of our questions, i.e., on an understanding of the so-called "price of locality".

more

Project A3: "The market for services, incentives, Algorithms, Implementation"

In Project A3 we describe and analyze the possible types of interaction between users, OTF providers, and suppliers of IT services. With the help of specific approaches and methods from the (algorithmic) game theory and information economics, we want to analyze and evaluate the effects of different incentive mechanisms. The goal is to minimize information asymmetries (e.g., regarding the quality of services) between the actors and the corresponding inefficiencies. In this context, we have to elaborate on the importance and role of guarantees, trust, and reputation as "information surrogates".

more

Project C2: "On-The-Fly Compute Centers"

In Project C2 methods for the timely and efficient execution of configured IT services are examined. We deal in particular with models and algorithms to exploit heterogeneity at different levels in OTF Compute Centers. This includes computing nodes that can consist of many different CPUs, GPUs, or FPGAs, networks of inhomogeneous computing clusters, and distributed single computers that are released by their owners. We consider primarily the development and analysis of scheduling strategies, which make most efficient use of the available heterogeneous computing resources and enable a profitable operation of the OTF Compute Centers.

more