enhanced efficiency of mapping distribution protocols in
play

Enhanced Efficiency of Mapping Distribution Protocols in Scalable - PowerPoint PPT Presentation

Enhanced Efficiency of Mapping Distribution Protocols in Scalable Routing and Addressing Architectures K. Sriram, Patrick Gleichmann, Young-Tak Kim, and Doug Montgomery July 2010 National Institute of Standards and Technology National


  1. Enhanced Efficiency of Mapping Distribution Protocols in Scalable Routing and Addressing Architectures K. Sriram, Patrick Gleichmann, Young-Tak Kim, and Doug Montgomery July 2010 National Institute of Standards and Technology National Institute of Standards and Technology Contact: Contact: ksriram@nist.gov ksriram@nist.gov Acknowledgement: Thanks to Dino Farinacci, Lixia Zhang, Joel Halpern, and Robin Whittle for their helpful comments and suggestions. This research was supported by the Department of Homeland Security under the Secure Protocols for the Routing Infrastructure (SPRI) program and the NIST Information Technology Laboratory Cyber and Network Security Program. 1

  2. Background • This work was originally presented in RRG in July 2007 at the Dublin IETF Meeting. • A revised version was presented at the IETF LISP WG meeting in March 2010. This current version reflects revisions based on feedback from that meeting. • Slides 6, 8, 9, and 14-17 have new or significantly revised material. • Detailed updated document is at: http://www.antd.nist.gov/~ksriram/EEMDP_ICCCN2010.pdf 2

  3. Overview of Map and Encap Solution ILM = ID Locator Mapping server Push ILM-R = ILM-Regional ILM Pull ILM Push Mapping ILM Response Distribution Pull ILM Query Push Network ILM-R ILM-R Control Plane query (ID/Loc registration (Signaling) mapping info of (ID/Loc request, response) mapping info) BGP (core Internet) IP-over-IP (based on locator) TR map-n-encap TR ETR1 ID/Loc cache ITR1 Forwarding mapper Plane Delivery networks Source AS1 Destination AS4 (based on ID) Network Network 3

  4. Managing Holes in Maps (Preview) Push a 7 /24 Pull a 14 /24 ILM = ID Locator Mapping server ILM-R = ILM-Regional ETR3 ETR2 ILM Map: (a 7 /24, ETR3) Map: (a 14 /24, ETR2) ILM (a/20, ETR1) ILM Response: (a 14 /24, ETR2) (a 7 /24, ETR3) ILM 1 Query: a 6 ILM-R Response: ILM-R (a/20, ETR1, MS=1) 1 Query: a 6 Map: (a/20, ETR1, Exceptions: a 7 /24, a 14 /24) ITR1 ETR1 MS = 1 indicates Packet destined to an existence of More 1 in a 6 /24 address a 6 Parent ETR for a/20 Specific(s) of this prefix at different ETR(s). [a 0 /24, … , a 6 /24, a 8 /24, … a 13 /24, a 15 /24] (Note: Most of a/20 is here; except subprefixes a 7 /24 and a 14 /24 which are elsewhere) 4

  5. Real-World Example: Hole in a PI Address Announced in BGP ‐ 4: Aggregate 129.6.0.0/17 Origin: AS49 More Specific 129.6.112.0/24 Origin: AS10886 EID to Locator Mapping: EID: ETR (equivalent) 129.6.112.0/24 ETR10886 129.6.113.0/24 ETR49 129.6.114.0/23 ETR49 129.6.116.0/22 ETR49 129.6.120.0/21 ETR49 129.6.96.0/20 ETR49 129.6.64.0/19 ETR49 129.6.0.0/18 ETR49 5

  6. Proliferation of Map Entries 129.6.0.0/17 129.6.0.0/18 129.6.64.0/18 ETR 49 129.6.64.0/19 129.6.96.0/19 ETR 49 Aggregate 129.6.96.0/20 Hole / Exception 129.6.112.0/20 ETR 49 129.6.120.0/21 129.6.112.0/21 ETR 49 Illustration of extra 129.6.116.0/22 129.6.112.0/22 set of map entries ETR 49 that become 129.6.112.0/23 necessary due to a 129.6.114.0/23 ETR 49 hole (in map and encap protocols) 129.6.112.0/24 129.6.113.0/24 (w/o the proposed ETR 10886 ETR 49 EEMDP solution) 6

  7. Measurement of # Prefix Holes 100000 # Subprefixes of Length x such that Each has a Less Specific with a Different Origin AS as 10000 Compared to the Subprefix 1000 Total # holes = 100 60988 10 1 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 32 Subprefix Length (x) Based on Routeviews RIBs 7 trace data – Feb 2010

  8. Avg. Map Multiplication Factor Due to Holes 16 Avg. Map Multiplication Factor due to 14 Holes of Subprefix Length x 12 Overall 10 avg. = 9.37 8 6 (w/o the proposed 4 EEMDP solution) 2 0 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 32 Subprefix Length (x) Based on Routeviews RIBs 8 trace data – Feb 2010

  9. Measurement of Proliferation of Maps 1000000 (Due to Effects of Holes of Subprefix Length x) Total # 100000 Extra # Maps in Database Maps in 10000 Database = 510508 1000 (Approx.) 100 (w/o the 10 proposed EEMDP 1 solution) 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 32 Subprefix Length (x) Based on Routeviews RIBs 9 trace data – Feb 2010

  10. Details of the Proposed Algorithm: Enhanced Efficiency of Mapping Distribution Protocols (EEMDP) 10

  11. Case 1: More-Specifics (Holes) Absent ILM = ID Locator Mapping server ILM-R = ILM-Regional ETR3 ETR2 ILM ILM ILM Response: (a/20, ETR1) ILM 1 Query: a 6 ILM-R MS = 00 ILM-R Response: (a/20, ETR1) 1 Query: a 6 ITR1 ETR1 Packet destined to an 1 in a 6 /24 address a 6 a/20 is entirely here; no except subprefixes 11

  12. Case 2: All More-Specifics Communicated a 7 /24 a 14 /24 ILM = ID Locator Mapping server ILM-R = ILM-Regional ETR3 ETR2 ILM ILM (a/20, ETR1) ILM Response: (a 14 /24, ETR2) (a 7 /24, ETR3) ILM 1 Query: a 6 ILM-R MS = 01 Response: ILM-R (a/20, ETR1) 1 Query: a 6 (a 14 /24, ETR2) (a 7 /24, ETR3) ITR1 ETR1 Packet destined to an 1 in a 6 /24 address a 6 Parent ETR for a/20 [a 0 /24, … , a 6 /24, a 8 /24, … a 13 /24, a 15 /24] (Note: Most of a/20 is here; except subprefixes a 7 /24 and a 14 /24 which are elsewhere) 12

  13. Case 3: Exception More-Specific Communicated without ETR Info (Lots of Mobile Nodes) a 14 /24 ILM = ID Locator Mapping server ILM-R = ILM-Regional ETR2 ILM EIDs from a 7 /24 ILM (a/20, ETR1) ILM are assigned to Response: (a 14 /24, ETR2) mobile nodes (a 7 /24, RR) ILM 1 Query: a 6 which are homed ILM-R MS = 10 to different ETRs Response: ILM-R (a/20, ETR1) 1 Query: a 6 (a 14 /24, ETR2) (a 7 /24, RR) ITR1 a 7 /24 ETR1 Packet destined to an 1 in a 6 /24 address a 6 Parent ETR for a/20 ETR = RR means Re- Request (i.e., send [a 0 /24, … , a 6 /24, a 8 /24, … a 13 /24, a 15 /24] map request) if destn. (Note: Most of a/20 is here; except subprefixes EID in that more- specific prefix a 7 /24 and a 14 /24 which are elsewhere) 13

  14. Case 4: Prioritized Subset of Maps for Exception More- Specifics Are Communicated a 7 /24 a 1 /24 a 12 /24 a 14 /24 ETR3 ETR4 ETR2 ILM Map: (a 7 /24, ETR3) Map: (a 14 /24, ETR2) ILM (a/20, ETR1) ILM Response: (a 14 /24, ETR2) (a 7 /24, ETR3) ILM 1 Query: a 6 ILM-R Response: ILM-R (a/20, ETR1, MS=11) 1 Query: a 6 Map: (a/20, ETR1, (a 1 /24, ETR4) Exceptions: a 7 /24, a 14 /24) ITR1 ETR1 MS = 11 indicates Packet destined to an existence of More 1 in a 6 /24 address a 6 Parent ETR for a/20 Specific(s) of this prefix at different ETR(s), and that only a prioritized [a 0 /24, a 2 /24, … , a 6 /24, a 8 /24, … a 11 /24, a 13 /24 a 15 /24] subset of those are (Note: Most of a/20 is here; except subprefixes a 7 /24 included in this map and a 14 /24 which are elsewhere) message. 14

  15. Conceptual Format for the Enhanced Map Response More Specific More Specific More Specific .... Prefix ETR MS K NE Map 1 Map 2 Map K MS = More Specific indicator K = # Maps to follow NE = Number of Exceptions ( NE > K) If for a more specific prefix, ETR = RR, it means ITR needs to query (Re-Request) for destination EID in that more-specific prefix 15

  16. Algorithm Description More # Exception Total # Interpretation Specific Maps Exceptions Indicator Included (K) (NE) (MS) 00 0 0 Map response has no exceptions. 01 k n e = k Map response has exceptions; All k map responses for the exception subnets are included. 10 k n e = k Map response has exceptions; All k map responses for the exception subnets are included but the ETR information for one or more specific subnets is “Re-request”; Subnets are further split into micro-subnets (e.g., mobile devices homed to different ETRs). 11 k (k < n e ) n e Map response has exceptions; # Exceptions exceeds threshold (H); Only a subset of exception maps is included; Maps for prioritized (frequently requested) subset of more specifics are included. 16

  17. Comparison of Max # Map Responses Attributable to Holes w/o and with EEMDP Reduction achieved with EEMDP = 90% 600000 Communicated due to Holes Max # Map Responses 500000 400000 300000 200000 100000 0 Map-n-Encap w/o EEMDP Map-n-Encap with EEMDP Based on Routeviews RIBs 17 trace data – Feb 2010

  18. Endpoint ID Aggregation at ETRs Map: (x/22, ETR7) Map: (y/22, ETR8) Map: (y 0 /24, ETR7, backup = yes) Map: (x 3 /24, ETR8, backup = yes) Map: (z 1 /24, ETR2) Map: (z 2 /24, ETR4) ILM-R ETR7 ETR8 Multi-homing ETR1 ETR6 ETR2 y 0 /24 ETR5 x 0 /24 y 3 /24 ETR3 x 3 /24 ETR4 y 2 /24 x 1 /24 x 2 /24 y 1 /24 z 2 /24 z 1 /24 18

  19. Conclusions and Future Work • Holes in ID-to-locator maps cause undesirable map proliferations • Significant reduction in map entries and map query/response traffic load is possible with the proposed EEMDP scheme • Substantial reduction in load on ITR’s memory and processor • More accurate quantification of benefits can be performed • Also introduced the notion of a loose hierarchy of ETRs with the potential benefit of aggregation of their EID address spaces 19

Recommend


More recommend