Wresting Control from BGP: Scalable Fine-grained Route Control UCSD / AT&T Research Usenix —June 22, 2007 Dan Pei, Tom Scholl, Aman Shaikh, Alex C. Snoeren,

Slides:



Advertisements
Similar presentations
Multihoming and Multi-path Routing
Advertisements

Multihoming and Multi-path Routing
Multihoming and Multi-path Routing CS 7260 Nick Feamster January
Internet Routing (COS 598A) Today: Topology Size Jennifer Rexford Tuesdays/Thursdays 11:00am-12:20pm.
© J. Liebeherr, All rights reserved 1 Border Gateway Protocol This lecture is largely based on a BGP tutorial by T. Griffin from AT&T Research.
1 BGP Anomaly Detection in an ISP Jian Wu (U. Michigan) Z. Morley Mao (U. Michigan) Jennifer Rexford (Princeton) Jia Wang (AT&T Labs)
1 Interdomain Routing Protocols. 2 Autonomous Systems An autonomous system (AS) is a region of the Internet that is administered by a single entity and.
How to Construct a Correct and Scalable iBGP Configuration Mythili Vutukuru Joint work with Paul Valiant, Swastik Kopparty and Hari Balakrishnan.
1 Route Control Platform – IEEE CCW 2004 Route Control Platform Making an AS look and act like one router Aman Shaikh AT&T Labs - Research IEEE CCW 2004.
1 Finding a Needle in a Haystack: Pinpointing Significant BGP Routing Changes in an IP Network Jian Wu (University of Michigan) Z. Morley Mao (University.
Traffic Engineering With Traditional IP Routing Protocols
1 Route Control Platform Making the Network Act Like One Big Router Jennifer Rexford Princeton University
Mini Introduction to BGP Michalis Faloutsos. What Is BGP?  Border Gateway Protocol BGP-4  The de-facto interdomain routing protocol  BGP enables policy.
Traffic Engineering in IP Networks Jennifer Rexford Computer Science Department Princeton University; Princeton, NJ
A Routing Control Platform for Managing IP Networks Jennifer Rexford Computer Science Department Princeton University
Slide -1- February, 2006 Interdomain Routing Gordon Wilfong Distinguished Member of Technical Staff Algorithms Research Department Mathematical and Algorithmic.
Dynamics of Hot-Potato Routing in IP Networks Renata Teixeira (UC San Diego) with Aman Shaikh (AT&T), Tim Griffin(Intel),
1 Route Control Platform – IEEE CCW 2004 Route Control Platform Making an AS look and act like a router Aman Shaikh AT&T Labs - Research IEEE CCW 2004.
1 Design and implementation of a Routing Control Platform Matthew Caesar, Donald Caldwell, Nick Feamster, Jennifer Rexford, Aman Shaikh, Jacobus van der.
A Routing Control Platform for Managing IP Networks Jennifer Rexford Princeton University
Internet Routing (COS 598A) Today: Telling Routers What to Do Jennifer Rexford Tuesdays/Thursdays.
Network Monitoring for Internet Traffic Engineering Jennifer Rexford AT&T Labs – Research Florham Park, NJ 07932
A Routing Control Platform for Managing IP Networks Jennifer Rexford Princeton University
Tesseract A 4D Network Control Plane
Backbone Networks Jennifer Rexford COS 461: Computer Networks Lectures: MW 10-10:50am in Architecture N101
A Routing Control Platform for Managing IP Networks Jennifer Rexford Princeton University
Hot Potatoes Heat Up BGP Routing Jennifer Rexford AT&T Labs—Research Joint work with Renata Teixeira, Aman Shaikh, and.
Fundamentals of Networking Discovery 2, Chapter 6 Routing.
Path-Vector Contract Routing Hasan T. Karaoglu, Murat Yuksel University of Nevada, Reno ICC’12 NGNI, Toronto June, 2012.
1 Meeyoung Cha (KAIST) Sue Moon (KAIST) Chong-Dae Park (KAIST) Aman Shaikh (AT&T Labs – Research) IEEE INFOCOM 2005 Poster Session Positioning Relay Nodes.
Network Sensitivity to Hot-Potato Disruptions Renata Teixeira (UC San Diego) with Aman Shaikh (AT&T), Tim Griffin(Intel),
1 Meeyoung Cha, Sue Moon, Chong-Dae Park Aman Shaikh Placing Relay Nodes for Intra-Domain Path Diversity To appear in IEEE INFOCOM 2006.
Authors Renata Teixeira, Aman Shaikh and Jennifer Rexford(AT&T), Tim Griffin(Intel) Presenter : Farrukh Shahzad.
I-4 routing scalability Taekyoung Kwon Some slides are from Geoff Huston, Michalis Faloutsos, Paul Barford, Jim Kurose, Paul Francis, and Jennifer Rexford.
9/15/2015CS622 - MIRO Presentation1 Wen Xu and Jennifer Rexford Department of Computer Science Princeton University Chuck Short CS622 Dr. C. Edward Chow.
CS 3700 Networks and Distributed Systems Inter Domain Routing (It’s all about the Money) Revised 8/20/15.
Dongkee LEE 1 BorderGuard: Detecting Cold Potatoes from Peers Nick Feamster, et al.
Using Measurement Data to Construct a Network-Wide View Jennifer Rexford AT&T Labs—Research Florham Park, NJ
Jennifer Rexford Fall 2014 (TTh 3:00-4:20 in CS 105) COS 561: Advanced Computer Networks BGP.
A Firewall for Routers: Protecting Against Routing Misbehavior1 June 26, A Firewall for Routers: Protecting Against Routing Misbehavior Jia Wang.
Vytautas Valancius, Nick Feamster, Akihiro Nakao, and Jennifer Rexford.
How to Construct a Correct and Scalable iBGP Configuration Mythili Vutukuru Joint work with Paul Valiant, Swastik Kopparty and Hari Balakrishnan.
Evolving Toward a Self-Managing Network Jennifer Rexford Princeton University
Yaping Zhu with: Jennifer Rexford (Princeton University) Aman Shaikh and Subhabrata Sen (ATT Research) Route Oracle: Where Have.
Evolving Toward a Self-Managing Network Jennifer Rexford Princeton University
The New Policy for Enterprise Networking Robert Bays Chief Scientist June 2002.
1 Agenda for Today’s Lecture The rationale for BGP’s design –What is interdomain routing and why do we need it? –Why does BGP look the way it does? How.
© 2005 Cisco Systems, Inc. All rights reserved. BGP v3.2—6-1 Scaling Service Provider Networks Scaling IGP and BGP in Service Provider Networks.
1 Chapter 4: Internetworking (IP Routing) Dr. Rocky K. C. Chang 16 March 2004.
Michael Schapira, Princeton University Fall 2010 (TTh 1:30-2:50 in COS 302) COS 561: Advanced Computer Networks
Internet Traffic Engineering Motivation: –The Fish problem, congested links. –Two properties of IP routing Destination based Local optimization TE: optimizing.
Doing Don’ts: Modifying BGP Attributes within an Autonomous System Luca Cittadini, Stefano Vissicchio, Giuseppe Di Battista Università degli Studi RomaTre.
CS 3700 Networks and Distributed Systems
CS 3700 Networks and Distributed Systems
Jian Wu (University of Michigan)
Border Gateway Protocol
Shedding Light on the Glue Logic of the Internet Routing Architecture
COS 561: Advanced Computer Networks
BGP supplement Abhigyan Sharma.
Interdomain Traffic Engineering with BGP
Wresting Control from BGP: Scalable Fine-grained Route Control
Internet Routing (COS 598A) Today: Topology Size
COS 561: Advanced Computer Networks
Routers Routing algorithms
COS 561: Advanced Computer Networks
COS 561: Advanced Computer Networks
Backbone Networks Mike Freedman COS 461: Computer Networks
Administrivia Paper assignments for reviews 2 and 3 are out
BGP Instability Jennifer Rexford
Computer Networks Protocols
Presentation transcript:

Wresting Control from BGP: Scalable Fine-grained Route Control UCSD / AT&T Research Usenix —June 22, 2007 Dan Pei, Tom Scholl, Aman Shaikh, Alex C. Snoeren, Kobus van der Merwe Patrick Verkaik

What is route control? ISP Route control ? ? Traffic destination Traffic ISPs need to control selection of alternate paths … in response to dynamic network conditions ISPs need to control selection of alternate paths … in response to dynamic network conditions measurement Historically routing provides simple connectivity But demands are changing: gaming, DDoS defense Historically routing provides simple connectivity But demands are changing: gaming, DDoS defense However, demands are changing:

IRSCP: enabling route control IRSCP: Intelligent Route Service Control Point –Next step after RCP [Caesar et al.] Enable route control for inter-domain traffic : –Automated, based on network conditions –Scalable to demands of large ISP –No changes to existing ISP infrastructure

Simple example: load balancing Customer ISP Traffic Egress router R1Ingress router R5 Border Gateway Protocol (BGP) BGP defaults to hot potato routing Effectively: shortest path routing Overloaded link Traffic on customer access links is balanced R3 R2 R4 Simple route control objective Customers are asking for this Yet BGP can’t do it! Simple route control objective Customers are asking for this Yet BGP can’t do it!

Customer 1 Why does BGP do that? ISP 1 ISP 2 Customer 2 External BGP (eBGP) session Internal BGP (iBGP) session ISP router External router Selec t route dest d R1 R3 R2 R4 Hundreds of ISP routers making local decisions Hot potato default policy: –Ensures consistency –..but precludes control over routing Hundreds of ISP routers making local decisions Hot potato default policy: –Ensures consistency –..but precludes control over routing

IRSCP route control Abstraction: “for this destination, direct traffic from this ISP router to that ISP router” Automated, based on network conditions: route control application Backwards compatible, consistent and scalable Selective: allow default BGP decision where desired

Outline IRSCP architecture Routing consistency Implementation Evaluation

Customer 1 Route control abstraction ISP 1 ISP 2 Customer 2 Traffic Egress links RC App iBGP session RC App IRSCP Assign- ments RC App assigns egress links to ISP routers Speak BGP to routers …using IRSCP Assign Route ? ? R4 R1 R3 R2

Route failover What if a route for egress link fails? –RC application at relatively slow timescale –IRSCP must fail over instantly So application sends a list of egress links for each ISP router –We call this a ranking

Customer 1 Rankings ISP 1 ISP 2 Customer 2 Traffic RC App IRSCP Rankings R1 R4 R3 R R4 R3R1 Select route for R1 and R3 Select route for R2 and R4 Rankings map traffic from ingress to egress arbitrarily And allow route fail-over at routing time-scale Rankings map traffic from ingress to egress arbitrarily And allow route fail-over at routing time-scale

What about scalability? IRSCP talks to many thousands of routers Responsible for route decision for each ISP router: –Computation –Single point of failure Maintaining BGP session for each router : –State per session –Each eBGP session may add a route 2-3 GB sufficient Distributed IRSCP

Customer 1 Distributed IRSCP ISP 1 ISP 2 Customer 2 Multiple IRSCP servers: –To distribute BGP sessions –For geographic diversity Routers may peer with several IRSCP servers IRSCP servers are replicas: exchange all routes IRSCP IRSCP session R4 R1 R3 R2

Outline IRSCP architecture Routing consistency Implementation Evaluation

Consistency Forwarding anomalies: Deflection Traffic R1 R2 R3 R2 Looping R1 R3 Rankings must be “consistent”

R1 R3 R2 R4 Example of inconsistent rankings RC application checks consistency constraints on rankings No anomalies Traffic R1 R2 Rankings R3R4 Deflection

Outline IRSCP architecture Routing consistency Implementation Evaluation

Expl. ranked decision process Hot potato/BGP decision process Routing information baseRoutes Prototype implementation IRSCP server RC App Rankings R1 IRSCP server R2 IRSCP server Import policy Export policy I1 I2

Outline IRSCP architecture Routing consistency Implementation Evaluation –Can IRSCP handle routing load in a real ISP? –Both explicitly ranked and hot potato decision process

Methodology Emulation of realistic large ISP topology and routing load Connect IRSCP implementation to emulation of ISP Measure if our implementation handles the emulated load

Customer 1ISP 2 Deployment Scenario IRSCP 240 external routers per POP 15 ISP routers per POP 40 POPs POP 1 IRSCP server per POP IRSCP session BGP session Throughput of IRSCP server depends on how many of each kind of session it has

Finding maximum throughput IRSCP Input update rate Route update receiver Mix of BGP and IRSCP sessions based on ISP scenario IRSCP implementation: –3.6-GHz Xeon –4 GB memory Search for maximum sustainable input rate: –Gradually increase input rate, sustaining for twenty minutes –Compute expected output rate given input rate –Once measured output rate falls behind, we’ve reached maximum throughput Output update rate Route update generator Multiplier

Maximum throughput 3600 Output rate Input rate Hot potato Explicitly ranked % in real ISP Average in real ISP 220 updates/s Flexibility of rankings has a cost But IRSCP handles Tier-1 ISP routing load, and more Flexibility of rankings has a cost But IRSCP handles Tier-1 ISP routing load, and more

Conclusion IRSCP route control platform: –Feedback of network conditions into route selection –Scalable, robust against failures, backward compatible –Powerful, yet safe ranking abstraction Enables new class of “route control application”: –Security –Traffic engineering –Customer-oriented services Trials of IRSCP for several applications taking place in AT&T!

Alex Aman Questions? Kobus Tom Dan Patrick

BGP sessions As we saw, IRSCP speaks iBGP to ISP routers For full control, IRSCP must also speak eBGP to external routers IRSCP Multihop eBGP session iBGP session I E1 E2 eBGP session Selec t route Select route for I

Example consistency constraint If e 1 outranks e 2 at R1 then must also do so at router(e 1 )=R3 Two simple consistency constraints Checked by RC application before sending to IRSCP Example: R1 R3 e1e1 e2e2 e1e1 e2e2 R1 R2 R3R4 e1e1 e2e2 R2 R4

Throughput Achieved output rate Estimated 95 perc. required input rate Achieved input rate Out of 255 routers per POP