CRIO: Scaling IP Routing with the Core Router-Integrated Overlay

1
CRIO Scaling IP Routing with the Core
Router-Integrated Overlay

• Xinyang Zhang Paul Francis Jia Wang
  Kaoru Yoshida

2
Outline

• CRIO Introduction
• CRIO Architecture
• CRIO Evaluation
• Related Works Conclusion

3
CRIO Introduction

• The Scaling Problem
• CRIO Approach
• Tunneling
• Virtual Prefix

4
The Scaling Problem

• A glimpse of current routing system
• Static table size
• Global IPv4 200K entries
• VPN 800K entries
• And more routes are coming IPV6,
  traffic-engineered, etc.
• Routing Dynamics
• BGP update churns
• Persistent instabilities
• Long convergence time

This talk is about the static characteristics of
the scaling Validity of CRIO approach

• Looking into the future
• Can we support a routing table twice (or 10
  times) the size of today?
• Can we rely on the hardware advances to alleviate
  the scaling pressure?

5
CRIO Approach

• Tunneling
• Revisit old idea (1996 by Deering, original by
  Bob Hinden and Robert Elz)
• Decouples addressing from topology
• Virtual Prefix
• Novel approach
• Greatly shrink forwarding table

6
Tunneling
Prefix TE Source
Mapping Adv. 24.1.1.0/24 TEPE2
Prefix TE Source
PE2 ---- BGP
24.1.1.0/24 ---- BGP/OSPF
24.1.1.0/24 PE2 Mapping
24.1.1.0/24 PE3 Mapping
Provider Networks
PE2
PE1
24.1.1.1
PE2
24.1.1.1
PE3
24.1.1.1
Routing Adv. 24.1.1.0/24 NHCE2
CE2
CE1
Customer Site C2 24.1.1.0/24
Customer Site C1
7
Tunneling Benefits

• Separate Mapping from Routing
• BGP only computes routes to TE prefixes
• On the order of one thousand entries
• Stable ISP provisioned prefixes
• Mappings are easy to distribute
• A mapping entry is the same no matter where it
  appears
• Support multi-homing without burdening the
  routing system

8
Forwarding Table Problems

• CRIO tunneling can not shrink forwarding
  information
• Forwarding table is expected to get larger
• Since CRIO supports for fine-grained multi-homing
• Benefits for having small forwarding tables
• Smaller memory requirement on routers line cards
• Faster transfer for forwarding table updates

9
Virtual Prefix
Prefix TE Source
Routing Adv. 24.0.0.0/8
PE2 ---- BGP
24.1.1.0/24 PE2 Mapping
24.2.2.0/24 PE4 Mapping
PE3
24.1.1.1
PE2
PE2
24.1.1.1
24.1.1.0/24
24.1.1.1
Customer Site
PE1
CE2
Prefix TE Source
PE2 ---- BGP
PE3 ---- BGP
24.1.1.0/24 PE2 Mapping
24.0.0.0/8 ---- BGP
10
Virtual Prefix Tradeoff

• Virtual prefixes provide a tuning knob for the
  router
• trade-off forwarding table size for path length
• Per-prefix basis
• Its a good trade-off to make
• Few prefixes handle most traffic
• Routers could shed most of their prefixes with
  very little overall increase in traffic volume
• Save routers from handling large amount of
  mapping updates
• Virtual Prefix is particularly suitable for VPNs

11
CRIO Evaluation

• CRIO Static Analysis
• Data Collection
• Forwarding Table Content
• Virtual Prefix Placement Policy
• Simulation Results
• Global Internet
• VPN
• CRIO Dynamics

12
CRIO Static Analysis

• Evaluate the static performance of CRIO by
  simulation
• Table Size VS. Path Length
• Simulated both Global Internet and VPN
• Simulation tool C-BGP

13
Data Collection

• Global Internet
• Topology
• POP-level from RocketFuel
• 23 Tier-1 ISP, 1219 POPs, 4159 inter-POP links
• Mappings
• Derived ltprefix, TEgt mappings from RocketFuel raw
  traces
• Internet Traffic Matrices
• Prefix-level, across all POPs
• Use Netflow records from Tier-1 ISP backbone
• VPN
• Data is from a large VPN provider and one of its
  national-sized customers

14
Forwarding Table Content

• Direct Entries
• Virtual Prefix Entries
• Extra Path-Shrinking Entries

Routing Adv. 24.0.0.0/8
Prefix TE Source
PE2 ---- BGP
PE3
24.1.1.0/24 PE2 Mapping
24.1.1.0/24
Customer Site
PE1
CE2
Prefix TE Source
Prefix TE Source
PE2 ---- BGP
24.1.1.0/24 ---- BGP
PE3 ---- BGP
24.0.0.0/8 ---- BGP
24.1.1.0/24 PE2 Mapping
15
Virtual Prefix Placement Policy

• Inter-ISP (Random)
• Intra-ISP
• Intra-ISP shortest customer path

Routing Adv. 24.0.0.0/8
Routing Adv. 24.0.0.0/8
Provider 2
Provider 1
24.1.1.0/24
PE1
Customer Site
CE2
16
Results Global Internet

• Path-length vs. Table-size

• Virtual Prefix does increase the path length
• Average path length converges quickly as the
  path-shrinking entries increases
• Reduce FIB size by 3-5 times with very little
  path length penalty

Increase the percentage of shortest path
traffic by increasing of Path-Shrinking Entries

99 Traffic uses shortest path
17
Results VPN

• Hub-Spoke nature of VPN traffic exploits the
  tradeoff better
• Reduce table size by 10-20 times with very little
  path length penalty

Cumulative Distribution of PE Routers
PE Routers In Hub Sites
18
CRIO Dynamics

• Routing Policy BGP ? Mapping tables
• Distribute mapping is simpler
• Mapping entry it is the same anywhere
• CRIO affords two benefits to BGP operation
• Fewer entries
• More stable entries
• Mapping table distribution ways
• BGP new attributes
• A separate overlay infrastructure

19
CRIO Conclusion

• Conclusion
• Future Work

20
Conclusion

• CRIO is a new routing architecture, aimed to
  provide
• Scalable and stable core routing
• Reduce BGP RIB by two order of magnitude
• FIB size reduction
• Reduce FIB by one order of magnitude for global
  Internet, 10-20x for VPN

21
Future Work

• Design and implement the mapping distribution
  infrastructure
• Study the dynamics aspect of CRIO
• Study the security aspect of CRIO
• Explore the use of CRIO to provide traffic
  engineering for multi-homed site
• Address (??) new management challenges

22
Thanks!