SFB 901: On-The-Fly (OTF) Computing - Individualised IT-Services in Dynamic Markets

在2011年的CRC初步提案中，我们制定了IT服务市场的愿景，描述了提供此类服务的方法，这在当时是新的，在很大程度上，今天仍然是新的：“我们的实时计算愿景是提供IT服务，这些服务是在基本服务之外单独和自动配置的，这些服务可以在市场上交易，并且可以灵活组合。与此同时，我们的目标是组织市场，其参与者通过适当的企业行动维持一个充满活力的服务市场。“在第三个资助期，我们将扩大我们的工作方向，有助于接受即时计算，从而在科学和工业中可持续地建立我们的方法和技术。例如，这些方向包括解决方案质量高的便利可用性、资源效率高的执行、通过安全交易实现的可靠性、质量保证和用户反馈，以及确保一个可能得到适当管理的市场发挥作用。在该中心，计算机科学家和经济学家在计算机网络和分布式算法、安全性和密码学、软件工程和验证、配置和机器学习、计算机工程和高性能计算、微观经济学和博弈论、商业信息学和管理等研究领域开展合作。为了证明OTF计算方法的可行性和一致性，我们实现了OTF计算的概念验证，可以在OTF市场的典型场景中发挥作用。我们将使用两个应用场景来说明我们的工作——自动机器学习（AutoML）和自动创建问答服务（AutoQA）。最后，我们将推进我们在CRC之外的实时计算整合工作，并在与行业合作伙伴和转移项目的活动框架内传递CRC中获得的专业知识。

In the initial proposal for our CRC in 2011, we formulated a vision of markets for IT services that describes an approach to the provision of such services that was new at that time and, to a large extent, is new still today:"Our vision of On-The-Fly Computing is to provide IT services that are individually and automatically configured out of basic services, which are traded on markets and can be flexibly combined. At the same time, we aim to organize markets whose participants maintain a vibrant market for services through appropriate entrepreneurial action."In the third funding period, we will expand our work in directions that contribute to the acceptance of on-the-fly computing and thus sustainably establish our methods and techniques in science and industry. These directions include, for example, convenient usability with high solution quality, resource-efficient execution, reliability through secure transactions, quality guarantees and user feedback, as well as assurance of a functioning, possibly appropriately regulated market. In the CRC, computer scientists and economists work together with research foci such as computer networks and distributed algorithms, security and cryptography, software engineering and verification, configuration and machine learning, computer engineering and HPC, microeconomics and game theory, business informatics and management.In order to demonstrate the feasibility and consistency of the OTF Computing approach, we implement a proof-of-concept for OTF Computing that can play through typical scenarios of an OTF market. We will illustrate our work using two application scenarios - automated machine learning (AutoML) and automated creation of Question Answering Services (AutoQA). Finally, we will advance our work for the consolidation of On-The-Fly Computing beyond the CRC and pass on the expertise gained in the CRC within the framework of events with industry partners and transfer projects.