Enhancing IT and Telecom Collaboration Through Grid Standardization

1. Grid Standardization & ETSI (May 2006) B. Berde, Alcatel R & I

2. Strategic vision: unblock IT+Telecom • “Utility” services on a large scale, providing conjointly: • Network and resource-end system services (maybe co-optimized) • Automated service, connectivity, and resource on-demand • Internet achieves a critical mass of users • Preserving the current levelof infrastructure-based competition for operators and service providers • New Grid technology:Adapting to the overall complexity while democratizeGrid services

3. Overview Identifying requirements and defining specifications for next generation client-to-business (or business-to-business) communication Why: When 95% of the overall traffic is peer-to-peer (Direct Connect and Bit): • Traffic characterization per end-user (with the courtesy of Acreo, 2005)

4. Transition to Telecom + IT convergence Broadening the scale of current Grid environments: • Coping with the infinite amount of resources available through Internet Triple questioning: • How to set up accurately network resources, matching Grid applications’ requirements over the Internet • How to transform accessibility of resource-end systems into Internet ‘networked resources’ • How to consider eventually optimization of those resources for Grid applications - with inherent QoS assurance and service monitoring New ‘philosophy’ and thereby optimization methodologies are relevant: • Virtualization of the resource location • Tradeoff between network resource and resource-end system capacity/performance (CPU, storage, etc)

5. What is about current applications and signaling • Client-to-Business and Business-to-Business applications can be both session based and non–session based • Such applications include (but are not limited to): • Peer-to-peer applications • Grid computing • eCommerce, eGovernment etc • requiring low latency and highly variable bandwidth in time • Therefore, signaling is performed today (and often mixed between): • Application layer • Service delivery layer supported by service connectivity layer e.g., session initiation/control • And rely on a pre-engineered IP/MPLS network (no interaction with)

6. Existing solutions: ‘hosts drive the network’ • Network resource reservation: • Today corporate resources are accessible via dedicated networks rarely accessible from the Internet • Application level communication protocols uses often large middleware (even though standardized) • Dedicated point-to-point connections; networking issues such as dynamic routing and traffic engineering are no longer required • Communication between applications and network: • A (domain) central management instance, Bandwidth-Broker (BB) type solutions; however, there is no efficient means today to communicate between (distributed) applications and networks • Service Resource Broker (SRB) • But generally, a sort of QoS overlay atop the network infrastructure e.g., Infranet initiative

7. Transition: the problem space Existing solutions: • Communication at the applicative level - network seen as a black box Innovation for production networking: • Service adaptive and cooperative communication between hosts and the network - network is integrated with ‘hosts’ Main challenges: • Evolution from existing 1 to 2 • Adjoining virtualized access to resource end-systems

8. Transition: how to link application and network Need for end-to-end network resource provisioning • Mapping of Grid application/middleware requirements to the network layer at host – (vertical arrow) • Provide edge-to-edge network resource reservation using the CURRENT network layer capabilities - (horizontal arrows at the network layer) Innovation: • Service adaptive and cooperative host-to-host (end-to-end) resource reservation mechanism

9. Target result: VAR service (Virtual Access to Resource) = VARS networks Broadband Access network Corporate Network = Server parks Client METRO Network Broadband Access network Client Core network Enterprise network Virtualization

10. Positioning the overall technical framework Project activity focus • Key 1: Enable integration of custom business applications across networks • Key 2: Open new markets to Telcos TCP/UDP APP TCP/UDP IP Virtual accessibility

11. Scope of acitivities at ETSI • Define and specify various (end-to-end) signaling needs with the related horizontal and vertical interfaces between: • Application level • Service delivery platform e.g., middleware • Service connectivity level • Service stratum independent from any network-layer specific mechanism e.g., QoS • for seamless (collaborative) application communication and resource information handling with the service stratum • Outline some business aspects: interactions between client / network / service provider • Create momentum and propose standardization topics Service stratum

12. Conclusions • Business aspects: • Grid technologies do not generate money today • Grid over the Internet needs the support of Telecom operators • The business sustainability is only possible if making the evolution from enterprise and scientific Grid to wide-area “Utility” Grid - the next step • “Utility” vision for Grid (over Internet): • Not a single transposition of existing Grid environments into Internet networked ones • Not all about middleware • Serious networking and other problems arise - real research challenges • Standardization at ETSI - may play a crucial role in the Telecom and IT convergence

13. www.alcatel.com

14. Service signaling: session enabled host-network interaction APP APP SAL SAL Host-network communication though Session-layer signaling • Session-level signaling is intercepted at the edge SAL level allowing to communicate session requirements to the network • Session-level signaling indirectly uses network reserved resources (not only for network resources) => Embedded network intelligence (1) (3) (2) SAL SAL IP IP Network Network Network GMPLS GMPLS Transport Transport Transport Transport Transport