Dynamic Circuits and Commercial Implementations of Dynamic Circuits
Time 04/22/08 03:00PM-04:00PM
Application-level innovation and infrastructure-level innovation are known to propel one another in a virtuous pattern of feature push and pull. Computing Grids, Supercomputing Centers, large data warehouses breed new paradigms suitable for research, healthcare, military applications. Their value grows as they get interconnected by wide-scale infrastructure with deterministic high throughput and/or low latency. At the same time, healthcare applications such as streaming media and 3D imaging are seeing dramatic expansion and requirement for increased resolution. A single interactive video session can generate network traffic rates as high as 6 Gbps. Our philosophy is to present high-value applications with the opportunity to directly drive a set of virtualized network resources within a range of policy-sanctioned behaviors. Through a mediation layer, the network?s ?knobs?, ?dials?, and ?levers? are abstracted and exposed to the applications in the form of virtual capabilities. The diversity of environments in which this approach may be used, and the resultant benefits, are best illustrated through representative use studies. Four such studies are therefore considered: Creation of an On-Demand service model; Introduction of Time-Value curves into Quality of Service based networks; Applying knowledge of healthcare workflows; Dynamic and Distributed Data Centers mobilize applications and virtual machines.
Radiology networks have an interesting challenge when interconnecting several hospitals in a rural region to provide a comprehensive diagnostic imaging service. The challenge arises due to several technological, workflow and organizational factors including: Continuous increase of image data generation; Increase in image size and number of images per case; More pervasive use of radiology images across the region by healthcare providers; Medical practice sub specialization that locate specialists in certain centers probably; and Ad-hoc, emergency retrievals that require immediate response. Consequences of a poorly designed and implemented network could cause delays or no access at all when healthcare providers try to retrieve medical cases resulting in limited diagnostic services to patients. The solution proposed in this paper is a new type of dynamic network that is workflow-aware, namely Workflow Engaged Network (WEN), can greatly improve MAN network performance by ensuring that the critical image transfers experience minimal delay. Our analysis shows that a 155 Mb/s MAN with WEN performs virtually equal to a non-WEN enabled 622 Mb/s MAN. Additionally, due to implementation of application-specific QoS capabilities and personalized SLA?s, this presentation proposes a re-optimization of other key resources such as storage and PACS systems for more effective and efficient workflows involving Radiologists. This will tend to relieve the shortage of such key personnel resources in rural areas.
Speaker Representative Nortel Networks
Speaker Bruce Schofield Nortel Networks
Speaker John Vollbrecht Internet2
Speaker Cees de Laat Universiteit van Amsterdam
Secondary tracks Leading Edge Networking: Planning and Engineering