Technical Reports

HPL-2012-111

Click here for full text: PDF

Cloud Networking: an Infrstructure Service Architecture for the Wide Area

Murray, Paul; Sefidcon, Azimeh; Steinert, Rebecca; Fusenig, Volker; Carapinha, Jorge
HP Laboratories

HPL-2012-111

Keyword(s): cloud computing; cloud networking; network virtualisation; infrastructure as a service; virtual infrastructure management; virtual infrastructure cloud networking, the provision of virtual infrastructure in a multi-administrative domain scenario, where data centre and network operators interact through defined interfaces to provide seamless virtual infrastructure. To support this scenario we introduce the flash network slice, dynamic elastic network connections that compose to form integrated cross-domain networks. Flash network slices support decomposition of virtual infrastructure into partitions that can be managed independently but seamlessly interconnected across administrative boundaries. The approach supports limited information disclosure about implementation details on behalf of the providers, scalability, heterogeneity, and a migration path from currently deployed infrastructure technologies to future network implementations. The resulting infrastructure services are suited to on- demand deployment of emerging cloud services such as content distribution, social networks and cloud-based IT applications.

Abstract: Next-generation printing service is an instance of the emerging class of cyber-physical systems; it is characterized by the large quantity of data that documents the factory operations. These data are ephemeral; the value of the data decays rapidly with time. The service solutions to serve this type of business must be scalable, provide high-throughput processing and low-latency (near real-time) analytical service, and cannot impose a prohibitive storage load. We describe a web service using provenance analytics engine as backend to aid and automate the print service management. It can leverage cloud computing infrastructure to ensure the scalability and multi- tenancy; adopts a novel use of component behavioral models, blends simulation and event management and querying, and provides multiple level of granularity through hierarchical process sequencing graph - all these contributes and ensures near real-time service delivery; uses component behavioral models to reproduce, rather than store, intermediate results * this can greatly reduce the storage demand as the intermediate content data in printing can be large; uses a non-annotation based approach eliminating the extra workload (computing & network bandwidth) that a growing header may impose to regular content data processing. Upon the instrumentation of this service, it can provide causality analysis & dependency analysis, replay ("roll back") and forecasting ("roll forward") service and real-time monitoring and alert.

8 Pages

Additional Publication Information: To be appeared in Future Network & Mobile Summit 2012: 21st Annual Conference

External Posting Date: May 21, 2012 [Fulltext]. Approved for External Publication
Internal Posting Date: May 21, 2012 [Fulltext]

Back to Index