1 / 13

TRILL Directory Assistance Mechanisms

TRILL Directory Assistance Mechanisms. draft -dunbar-trill-scheme-for-directory-assist- 06 draft-eastlake-trill-ia-appsubtlv - 03 Linda Dunbar, Donald Eastlake Radia Perlman, Igor Gashinsky , Yizhou Li Changes from presentation in July in Red. Goal.

annora
Download Presentation

TRILL Directory Assistance Mechanisms

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. TRILL Directory Assistance Mechanisms draft-dunbar-trill-scheme-for-directory-assist-06 draft-eastlake-trill-ia-appsubtlv-03 Linda Dunbar, Donald Eastlake Radia Perlman, Igor Gashinsky, Yizhou Li Changes from presentation in July in Red TRILL Directory Assist Mechanisms

  2. Goal • To reduce multi-destination traffic by • reducing or eliminating unknown unicast flooding • when appropriate, locally responding to ARP, ND, and RARP requests or converting them to unicast requests using address mapping directories. • An appropriate source of directory information is a Data Center orchestration system. TRILL Directory Assist Mechanisms

  3. Two Directory Types • Push Directories • Data and updates pushed out on a per Data Label (VLAN or Fine Grained Label) basis to all subscribing candidates. • Pull Directories • Each Pull Directory responds to requests in a set of Data Labels it advertises. TRILL Directory Assist Mechanisms

  4. Two Directory Types • Push Directories • Redundant push directories supported that can be configured as to how many should be active at once. • Data conflicts and which push directories should be active arbitrated by priority. • Push Directory state machine added. • Primary Push Directory can provide the data to secondary directories. TRILL Directory Assist Mechanisms

  5. Two Directory Types • Pull Directories • Data and negative responses include an expiry time for caching. • Unsolicited updates sent for unexpired cached information. TRILL Directory Assist Mechanisms

  6. Out of Scope • How directories are originally populated with information and updated as the network changes. • But orchestration systems seem like a good source. TRILL Directory Assist Mechanisms

  7. Push Directory Mechanisms • Uses the TRILL ESADI protocol for reliable Data Label scoped data flooding services. • draft-ietf-trill-esadi-03.txt (can be secured using ESADI authentication) • Client simply advertises (in core IS-IS) participation in ESADI for a Data Label and will be sent the data and updates. • Push Directories for a Data Label can see each other in that ESADI instance to arbitrate which should be active. TRILL Directory Assist Mechanisms

  8. Pull Directory Mechanisms • Uses the RBridge Channel mechanism for requests and responses. • draft-ietf-trill-rbridge-channel-08.txt (can be secured using draft-eastlake-trill-channel-tunnel) • Can optionally include the actual frame that caused the pull in pull request. • Can be hosted on an end stations, in which case the TRILL switch by which it is reachable proxies for it. TRILL Directory Assist Mechanisms

  9. Push Pull Policies • Clients can have a wide variety of policies: • Type of directory use: • Just use pushed data. • Just use pulled data. • Consult pushed and cached pull data and do a pull if no match found. • Behavior if no match • Either immediately or after failed pull: • Discard packet • Flood packet TRILL Directory Assist Mechanisms

  10. Push Pull Policies • Push Directories may be the most appropriate for Data Labels with a smaller number of end stations that mostly all talk to each other. • Pull Directories may be the most appropriate for Data Labels with larger numbers of end stations with sparse intercommunication. • If a Data Label has a few end stations everyone talks to, but otherwise has sparse intercommunication, you could push information for the few and use pull for the rest. • A client pulling could just pull from the nearest relevant pull directory or could pull from all revenant pull directories and use the first response it gets, etc. TRILL Directory Assist Mechanisms

  11. Added Material • Text on events that may cause directory use • Forged native frames • Unknown destination MAC • ARP, ND, or RARP • Optional Layer 3 address learning • Should this material be split off into a separate draft? TRILL Directory Assist Mechanisms

  12. Address Mapping Data Representation • Both push and pull directories use the “Interface Addresses” APPsub-TLV • draft-eastlake-ia-apsubtlv-03.txt • Provides for encoding a flexible set of addresses that all represent the same Interface (port). • For example { an IPv6 address, a 48-bit MAC address, a Data Label, a TRILL switch nickname }. • Could be used to look up the IPv6 address or the MAC address within the Data Label to get the other addresses. TRILL Directory Assist Mechanisms

  13. Next Steps • WG adoption of draft-dunbar-trill-scheme-for-directory-assist-06.txt and draft-eastlake-trill-ia-appsubtlv-03.txt • Solicit reviewers at this meeting to review after WG adoption. TRILL Directory Assist Mechanisms

More Related