Virtualised Infrastructures
Services available on the internet are running on a network of computers making use of particular devices and storage capability. Virtualisation technology makes it possible that companies roll out new services and applications without investing in and maintaining an expensive IT infrastructure.
In fact, through virtualisation, IT resources can be made available as a service (this is sometimes called Resource as a Service) and depending on the needs of the moment, for instance how many users are simultaneously accessing an application, the necessary amounts of resources can be used. Service Oriented Infrastructure or SOI is a system for describing and delivering IT infrastructures as a service.
Service Oriented Infrastructures
Service Oriented Infrastructures enable organizations to move from allocating dedicated resources to each application to dynamic resource allocation in which virtual processing, storage and network resources are assigned to the applications as needed.
By providing better resource utilization costs are reduced. Also the reliability of applications will be increased since in case of hardware failure other resources can take over easily. The infrastructure can automatically allocate additional resources in real time as an applicationâ??s workload increases. Thus, SOI delivers bottom-line benefits to the enterprise. A good working SOI is one of the essential elements of the Future Internet.
Virtualisation of resources
A service oriented infrastructure delivers infrastructures as a service from a pool of shared virtualized resources. Shared resources are aggregated, secured and presented as services across a network.
The SOI model consists of two parts: hardware resources and infrastructure services. Hardware resources include: network, storage, sensor networks and computing. Infrastructure services include: Security, data management services, computing services, directory, provisioning, capacity planning, fault monitoring, metering, and billing. To increase use and efficiency, the hardware resources need to be pooled and made dynamically available.
Infrastructure services help with the provisioning, monitoring, scaling, and secure operation of hardware resources. Although the most immediate motivations for virtualization of resources are improved resource utilization and lower costs, the ultimate goal is to use the abstraction between applications and the underlying resources to manage IT infrastructure as a service.
This provision of infrastructure services supporting different levels of abstraction and virtualisation of resources, dynamic monitoring of SLAs and the dynamic reallocation of services are at the core of the future research that is required to foster the Internet of Services.
The following projects are contributing to the Virtualised Infrastructures area.
RESERVOIR addresses the area of service-oriented infrastructures. It will develop and demonstrate an architecture and reference implementation for reliable and effective delivery of services as utilities. By harnessing virtualization and grid technologies across administrative domains, RESERVOIR will provide a foundation for a cost-competitive transparently and flexibly provisioned and managed like utilities. Instead of static over provisioning, RESERVOIR will allow for the nimble relocation of resources while at the same time ensuring Service Level Agreement (SLA) compliance and security guarantees.
IRMOS aims to design, develop, integrate and validate a service-oriented infrastructure that meets the requirements of interactive real-time applications. IRMOS will apply emerging real-time methodologies and technologies that are suitable for soft real-time systems in the context of SOIs, for the purpose of building foundations of the next-generation SOA-oriented interactive real-time applications, that will benefit of both the location independence and robustness typical of SOIs, and the predictability in timing behaviour (interactivity and response time, throughput) typical of soft real-time systems. IRMOS addresses the possible issues that are brought up when virtualization is used in the context of real-time applications.
SmartLM addresses the fact that traditional software licensing methods are not suitable when resources are shared, since the software is no longer tied to one particular computer. SmartLM will provide a new generic licensing virtualization framework based on standards, and integrate it in major Grid middleware solutions. The overall approach consists of treating and implementing software licenses as Grid services thus providing platform independent access just like other virtualized resource.
Licenses will become Grid services; a promising approach to overcome the limitations of current monolithic licensing models. Licenses will be managed as agreements, extending the conventional Service Level Agreements (SLAs) which are made today between sellers and buyers in the market. Licenses will be dynamic in order to support agreements that may change over time and where the dynamic negotiation between service provider and consumer is needed.
STREAM aims at producing a highly scalable infrastructure for processing in real time massive data streams such as the IP traffic of an organization, the output of a large sensor network, the e-mail processed by an ISP, the market feeds from stock exchange and financial markets, the calls in a telco operator, credit card payments, etc. STREAM aims at scaling system size by an order of magnitude, and providing unsupervised (autonomous) operation which will allow for much broader deployment of such data intensive services to areas that need to manipulate large amounts of information in a cost-effective manner. This will enable a myriad of new services and applications in the upcoming Internet of Services.
Closely linked to STREAM is the project ADMIRE. ADMIRE will accelerate access to data exploitation. It will achieve this by delivering consistent and easy-to-use technology for extracting information and knowledge. To cope with complexity, change and heterogeneity of services, data, and processes, an abstract view of data mining and integration will be provided. This will support users and developers of data mining and integration processes.
This page is maintained by: Sabine Posdziech