This document discusses multihoming and mobility support in different network architectures. It analyzes RINA, a clean-slate design, in comparison to LISP and Mobile IP. RINA uses late binding of node addresses to points of attachment, limiting the scope of location updates. Simulation results show RINA has lower packet delivery costs than LISP due to its routing approach. RINA inherently supports mobility as nodes can join and leave different layers of the network.

1. On the Cost of Supporting Multihoming
and Mobility
Ibrahim Matta
Computer Science
Boston University
Joint work with Vatche Ishakian, Joseph Akinwumi, John Day
I. Matta 1

2. Mobility = Dynamic Multihoming
 Hosts / ASes became increasingly multihomed
 Multihoming is a special case of mobility
 RINA (Recursive InterNetwork Architecture) is a
clean-slate design
 RINA routing is based on node addresses
µ Late binding of node address to point-of-attachment
 Compare to LISP (early binding) and Mobile-IP
 Average-case communication cost analysis
 Simulation over Internet-like topologies
I. Matta

3. Whatʼs wrong today?
Applications Web, email, ftp, … Applications
Transport Transport
Network Network Network
12
7.2
8.
Data Link DL DL Data Link
19
.12
7.
5 8.10
15
Physical PHY PHY Physical
.1
12
128.197.0.0 www.cs.bu.edu
128.10.0.0
128.197.15.10
 We exposed addresses to applications
 We named and addressed the wrong things

4. RINA offers better scoping
Applications Web, email, ftp, … Applications
Transport TCP, UDP, … Transport
IPC
Network IP Network Network
Data Link DL DL Data Link
IPC IPC
Physical PHY PHY Physical
 E2E (end-to-end principle) is not relevant
µ Each IPC layer provides service / QoS over its scope
 IPv6 is/was a waste of time!
µ We donʼt need too many addresses within an IPC layer

5. RINA: Good Addressing
Bob
want to send message to “Bob”
A “Bob”B B
To: B I1 I2
 Destination application is identified by “name”
 App name mapped to node name (address)
 Node addresses are private within IPC layer
µ Need a global namespace, but not address space
µ Destination application process is assigned a port number
dynamically
5

6. RINA: Good Addressing
Bob
want to send message to “Bob”
A B
To: B IPC processes
I1 I2 on same
machine
BI2
 Late binding of node name to a PoA address
 PoA address is “name” at the lower IPC level
 Node subscribes to different IPC layers
6

7. RINA: Good Routing
source destination
 Back to naming-addressing basics [Saltzer ʼ82]
µ Service name (location-independent) 
node name (location-dependent) 
PoA address (path-dependent)  path
 We clearly distinguish the last 2 mappings
 Route: sequence of node names (addresses)
 Map next-hopʼs node name to PoA at lower IPC level
I. Matta 7

8. Mobility is Inherent
MH CH
 Mobile joins new IPC layers and leaves old ones
 Local movement results in local routing updates
8

9. Mobility is Inherent
CH
 Mobile joins new IPC layers and leaves old ones
 Local movement results in local routing updates
9

10. Mobility is Inherent
CH
 Mobile joins new IPC layers and leaves old ones
 Local movement results in local routing updates
10

11. Compare to loc/id split (1)
 Basis of any solution to the multihoming issue
 Claim: the IP address semantics are overloaded as both
location and identifier
 LISP (Location ID Separation Protocol) ʻ06 EIDx -> EIDy
EIDx  EIDy RLOC1x  RLOC2y
EIDx  EIDy
Mapping: EIDy  RLOC2y
I. Matta

12. Compare to loc/id split (2)
 Ingress Border Router maps ID to loc, which is the
location of destination BR
 Problem: loc is path-dependent, does not name the
ultimate destination EIDx -> EIDy
RLOC1x RLOC2y
EIDx  EIDy
Mapping: EIDy  RLOC2y

13. LISP vs. RINA vs. …
 Total Cost per loc / interface change =
Cost of Loc / Routing Update +
ρ [Pcons*DeliveryCost + (1-Pcons)*InconsistencyCost]
ρ: expected packets per loc change
Pcons: probability of no loc change since last pkt delivery
 RINAʼs routing modeled over a binary tree of IPC layers:
update at top level involves route propagation over the
whole network diameter D; update at leaf involves route
propagation over D/2h, h is tree height
I. Matta

14. LISP
I. Matta

15. LISP
I. Matta

16. RINA
I. Matta

17. RINA
I. Matta

18. RINA
I. Matta

19. MobileIP
I. Matta

20. LISP vs. RINA vs. …
8x8 Grid Topology
RINA uses 5 IPC levels; on average, 3 levels get affected per move
LISP
RINA
I. Matta

21. Simulation: Packet Delivery Ratio
RINA
 BRITE
generated 2-
level topology
 Average path
length 14 hops
 Random walk
LISP
mobility model
I. Matta 21

22. Simulation: Packet Delay
LISP
RINA
I. Matta 22

23. Bottom Line: RINA is less costly
 RINA inherently limits the scope of
location update & inconsistency
 RINA uses “direct” routing to destination
node
 More work: prototyping
I. Matta

24. Thank You
Questions?
I. Matta