Nowadays, there is a multitude of computing resources readily available via the Internet. Cloud or grid infrastructures provide seemingly unlimited computing and storage capacities without interruption. The availability and scalability of cloud services has to be achieved by cloud providers, and consumers of cloud services expect that guaranteed levels of service quality (QoS) are fulfilled. Non-functional properties like availability, performance and the existence of adequate service level agreements (SLAs) have been recognized as key requirements for customers to move their IT and their business to the cloud .
by Frank Schulz, SAP Research Karlsruhe
From the providers’ point of view, the offered service qualities have to be achieved in an economically efficient way. In addition to leveraging economies of scale, resources need to be shared between customers in order to maintain a high degree of resource utilization. At the same time, the delivered services have to be isolated from each other not only in functional terms, but also with respect to non-functional properties. Given these requirements, the management of services during the complete life cycle creates complex challenges for cloud providers.
On the technical level, there exist basic quality metrics like availability or the configuration of (virtual) machines. On higher levels like platforms (PaaS) and applications (SaaS), business users ask for other QoS metrics which relate more directly to business requirements. This includes end-to-end guarantees like dialogue response time or completion time of batch jobs. Higher level services are realized by the combination of other services or components. This can be the direct call of web services, the usage of external services within workflows, or the non-technical bundling of several services into one offer. In all cases, a translation from technical low-level metrics to high-level business terms is required, combined with appropriate aggregation mechanisms.
A starting point for the management of complex services is the continuous monitoring of all relevant metrics and system parameters. Some of them are publicly monitored, like the availability of cloud APIs  , most others need to be assessed specifically for each case. Next, a scheme for the translation of these metrics into business-relevant terms is needed. Due to the complexity of systems and applications, such a scheme can often be established only by empirical measurements of the whole system under various workload and parameter conditions. Relating these technical considerations to business terms allows the derivation of risk assessments and business-related decision support for cloud providers. During runtime, the dynamic adjustment of services and infrastructures creates new possibilities for delivering the agreed quality of service while keeping an optimal level of resource utilization.
These challenges are addressed by the ValueGrids project  , which is supported by the German Federal Ministry of Education and Research (BMBF). It aims at bridging the gap between technology and business in order to facilitate the usage of cloud and grid infrastructures in complex business scenarios.
Image sources: http://www.wechselwild.com
Category: Guest Posts