Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Future Emergency Telecommunication Scenarios over the Internet Emergency Telecommunications Workshop 26’th-27’th, February 2002, ETSI, Sophia Antipolis, France Dr. Ken Carlberg [email protected] Outline Background The Challenge: Emergency Services over the Internet Services & Protocols Operational Scenarios Usage Scenarios Summary Background: Internet A network of networks Self autonomy Minimal requirements to be an ISP May use routing protocols May use non-FIFO queues No traffic engineering requirements Most congestion occurs at access points Tier-1 ISPs usually have high excess capacity at exchange points U.S. centric view at times Counter example includes Trans-Atlantic link(s) Background: Internet (2) Default service model is Best Effort “send and pray” No minimal level of QoS TCP is ~90-95% of traffic load Adaptive to congestion, but cost is degraded service Security is an issue with IP networks Denial of service: NIMDA, ICMP Echo Request, … Spoofing Challenge Distinguish emergency traffic Provide separate level of service Policy and/or regulatory issue Value added services Alternate path routing Interoperation with PSTN Mapping code points (at a minimum) Achieving the above with minimal changes to existing IP protocols Services & Protocols Past, Current, and/or On-Going (sample set) SIP/Megaco/H.323 MPLS Diff-serv Int-Serv/RSVP Instant Messaging & Presence Other WFQ RED Observations Existence of protocols does NOT equate to availability by vendors or service providers MPLS Local domain service Complex and possibly overkill for many ISPs Int-Serv/RSVP Market rejection of end-to-end model Diff-serv Local domain service Basic (AF) service can be accomplished with WFQ/RED So,….be a pessimist about what exists, and leverage what you can use VoIP with QoS: Near Term IP Backbone Single-Hop ISPs IP as a private network Single control of resources Telcos No Inter-ISP SLAs Single control of resources SS7 Evolution towards NGN SS7 Signaling VoIP (SIP/H.323 over IP) WFQ ISP Cloud Internet PSTN Stub IP Domain VoIP PSTN PBX Client IP Stub ISP Cloud Diff-Serv (AF) Client IP Stub ETS Operation: Near Term Label calls for ETS SIP Resource Field (draft) H.323 Priority Field (draft) Policy defines actions (part of SLA) Preemption / non-preemption Traffic engineered paths SLA’s dictate usage (e.g., diff-serv code points) e.g., #1) AF (Class 1) for Signaling, EF for VoIP e.g., #2) AF (Class 3) for Signaling & VoIP Access control at the edge ETS Operation: Mid Term Alternate path routing BGP could not support Emergency attribute Routers straining to support number of routes Convergence time problem Network View SIP Server TRIP View Application Layer routing e.g., Telephony Routing over IP (TRIP) TRIP Route ETS Operation: Mid Term (2) Admission Control Performed at edge of diffserv domain Admission Control Core/Internal congestion AF: use drop precedence EF: requires careful traffic engineering to avoid congestion Call/Data (1) Call/Data (2) (emergency) Potential augmentation New code point to distinguish emergency EF from normal EF Admission Control Call/Data (1) (emergency) Call/Data (2) ETS Usage Traveling Authentication & Capability Similar to GETS Non-ubiquitous service ETS “islands” connected via best effort service Goal is ever increasing wide spread support Payment and/or regulation are important issues because…. …..“There is no such thing as a free lunch” ETS Future? QoS Gateways Forward Error Correction (FEC) Redundant transmission Transcoding Semi-Active Networking Very leading edge approach E.g., Cisco Intelligence Engine 2100 XML/policy based control of network elements Negotiated service with user Degraded service if admission control fails Summary Autonomous & independent nature of IP networks makes support of ETS difficult Be pessimistic about available services “We” probably have 85% of what is needed to supporting ETS More options will exist for the ETS user