1 / 7

Multihoming: the SCTP perspective

Multihoming: the SCTP perspective. Multihoming IPv6 Working Group multi6 Lode.Coene@Siemen s . com. SCTP: Stream Control Transmission Protocol. Described in RFC 2960 and 3309 Multihomed host: endpoint with more than one IP address allocated to it.

williamdixon
Download Presentation

Multihoming: the SCTP perspective

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Multihoming: the SCTP perspective Multihoming IPv6 Working Group multi6 Lode.Coene@Siemens.com

  2. SCTP: Stream Control Transmission Protocol • Described in RFC 2960 and 3309 • Multihomed host: endpoint with more than one IP address allocated to it. • Each IP address represents a path through the network towards that endpoint and has separate congestion control variables • SCTP does NOT know if the paths are completely, partly or not distinct at all from each other. • Ideally: we would like all paths to be completely distinct • Practice: we simply don’t know... • Heartbeats for every path: keeps paths variables up-to-date and checks if the path is alive. Title presentation

  3. SCTP multihoming Title presentation

  4. SCTP: Stream Control Transmission Protocol • Multihoming did open some new horizons for us: • Add/remove IP address: in draft version <draft-ietf-tsvwg-addip-sctp-07.txt> • Based on add/remove IP, mobility support by the endpoint/host <draft-riegel-tuexen-mobile-sctp-02.txt>.. And some other drafts(got already out of hand I think...) • Loadsharing accross the different paths of a multihomed association: still a research issue, but it is certain that the endpoint has a better view of loadsharing than any node within the path ... • Proves that things that are done at the endpoint are really simplere and more powerfull than doing it in the network... OK, alllright ..The end-to-end principle rules...happy now... • In conjuction with partial-relialebility SCTP(PR-SCTP)... better quality video?!?!.. <draft-ietf-tsvwg-prsctp-00.txt> Title presentation

  5. SCTP: Stream Control Transmission Protocol • Some multihoming issues are documented in RFC 3257: • Some multihoming issues are documented in <draft-coene-sctp-multihome-04.txt> • Host needs routing table: no big deal for host with a limited number of associations/connections to different remote peers • Needs a big table in case of lots of associations/connections to lots of remote peers, but it is just throwing memory at the problem on the host, so this is no big deal either • How to fill in the routing table: implementation dependant: preconfigured,.. Caching when sending or receiving the INIT(=start of SCTP association)... Title presentation

  6. SCTP: Stream Control Transmission Protocol • Some multihoming and NAT interworking issues are documented in <draft-coene-sctp-multihome-04.txt> • Allright, I know this ain’t a NAT WG • But we put it in to be sure • We know in any case, trying to resolve that for the general case would require communication and coordination between the different NAT boxes anyway. And it looks like a very hard problem in the n*m range, so lets leave it there... • Observation: If multiple, distinct paths towards a remote destination had been in IP from the start on, NAT boxes wouldn’t have had a chance in Hell of being introduced or used in the network. (I should know, because in the network I do(SS7) I have yet to see the first NAT box being used) Title presentation

  7. Conclusion • View from the transport plane: • SCTP :we are getting along with the present solutions • TCP: see Christians draft <draft-huitema-multi6-experiment-00.txt> • It works... Title presentation

More Related