Potential Themes for research in SDN, NFV and Internet Architecture András Császár, Ericsson Research
Background Existing and Near Future Research SDN (logically) centralized control plane programmability for novel networking virtualization (abstraction) simplified network (re)configuration and policy enforcement NFV virtualizing servers and appliances that provide network functions by optimization and automation server costs can be reduced Key technology enablers towards cost reductions (in CAPEX and OPEX) and new business models Bigger focus in network service providers / operators than on end-users Ongoing activities work on unifying SDN and NFV E.g., FP7 UNIFY or Software Network strand of the 5GPPP pre-structuring model Will soon have flexibly programmable services including network, compute and storage have solutions for abstracted and programmable resources domains, stacked recursively, where the lowest level maybe a single node with its local resources Potential research work topics in SDN, NFV and Internet Architecture Public Ericsson AB 2014 2014-09-30 Page 2
Further Themes 1. New Applications Understand potential new applications and their new requirements Application layer is where innovation happens! E.g., Tactile Internet, Industrial Internet or industrial control applications The new SDN-NFV service platform is a good basis to support these E.g., latency requirements of Tactile-/Industrial-Internet by automatically placing servers as close to the subscribers as possible Get the innovation ecosystem bootstrapped Still need realisations of novel and sensible applications on the SDN+NFV service platforms Potential research work topics in SDN, NFV and Internet Architecture Public Ericsson AB 2014 2014-09-30 Page 3
Further Themes 2. SDN+NFV: On the Way to be the Basis of the Future Internet 1. Multi-technology network domains is being worked on Telecom market fragmented, especially the EU Multi-operator inter-connect to construct (e.g. Pan-European) services is needed Replacement of roaming in the flexibly programmable software networks world Multi-operator environment for control and orchestration East-west interfaces, and/or north-south interfaces between orchestrators? Protocol? What info is shared? What control and access can be allowed? Potential research work topics in SDN, NFV and Internet Architecture Public Ericsson AB 2014 2014-09-30 Page 4
Further Themes 3. SDN+NFV: On the Way to be the Basis of the Future Internet 2. A true end-to-end view requires still more Including and integrating resources in enterprise-/home-network premises Construct end-to-end end-user and enterprise services with a mix of resources from the customer and multiple operators What information is shared, who is responsible for what, what type of applications can benefit from such a case? Potential research work topics in SDN, NFV and Internet Architecture Public Ericsson AB 2014 2014-09-30 Page 5
Further Themes 4. Hardware-Defined Networking Still a challenge for SDN or in particular for OpenFlow: Flexibly programmable, low power, high performance, and yet cost efficient hardware Even more significant in OpenFlow 2.0 and similar initiatives (e.g., PIF/POF) Role of HW acceleration Role of purpose oriented chips, ASICs, NPUs, CPUs A different angle of HDN: optics in the network, in the DCs, in the boards, in the chips Optical Packet Switching research to be re-loaded? Potential research work topics in SDN, NFV and Internet Architecture Public Ericsson AB 2014 2014-09-30 Page 6
Further Themes 5. Platform Features of SDN&NFV Service Environment What features, properties can be offered by the platform in an app-independent manner? E.g., reliability, VM migration, dynamicity changes, etc.) What applications can rely on these built-in features? Needs detailed investigation and demonstrators Potential research work topics in SDN, NFV and Internet Architecture Public Ericsson AB 2014 2014-09-30 Page 7
Further Themes 6. End-to-End Encryption and the Role of NW Service Providers E2E encryption: a fact for most traffic What is the impact on the Internet architecture? Networks can hardly do any smart features with encrypted traffic Role of network service providers: plain bitpipe or is there more? We might reach the limits of over-provisioning at some point Network intelligence and smart network features could become more important Willingness to pay for quality (Qo*) Needed: Knowledge about type/characteristics of encrypted traffic But should not raise end-user privacy violation issues Operators allying and incentivizing content providers and application developers Potential research work topics in SDN, NFV and Internet Architecture Public Ericsson AB 2014 2014-09-30 Page 8
Further Themes 7. Application-Network Interface We may see interest for: Very dynamic Ad-hoc requests for Secure end-to-end connections with QoS guarantees for flows finer granularity than 5-tuples (e.g. flows multiplexed in higher layers) (Traffic type reveal) Application-network interface to request network services Software interfaces but at much finer granularity than SDN currently Potential research work topics in SDN, NFV and Internet Architecture Public Ericsson AB 2014 2014-09-30 Page 9
Highlights SDN & NFV Bootstrap SDN+NFV innovation ecosystem Realisations of novel and sensible applications on the SDN+NFV service platforms End-to-end Multi-operator inter-connect Including enterprise-/user-resources Hardware: programmable, high performance, low cost, low power End-to-end encryption: impacts and challenges Application-network interface for requesting advanced network services Potential research work topics in SDN, NFV and Internet Architecture Public Ericsson AB 2014 2014-09-30 Page 11