ROUTEVIEWS EVOLVES: Modernizing the BGP Collector for Today's - - PowerPoint PPT Presentation

▶
routeviews evolves modernizing the bgp collector for
SMART_READER_LITE
LIVE PREVIEW

ROUTEVIEWS EVOLVES: Modernizing the BGP Collector for Today's - - PowerPoint PPT Presentation

Mar 03, 2024 217 likes •644 views

ROUTEVIEWS EVOLVES: Modernizing the BGP Collector for Today's Researcher ROUTEVIEWS A collaborative router looking glass to share BGP views among network operators and researchers. ROUTEVIEWS ROUTEVIEWS A collaborative router looking glass

slide-1
SLIDE 1

ROUTEVIEWS EVOLVES: Modernizing the BGP Collector for Today's Researcher

slide-2
SLIDE 2

ROUTEVIEWS

ROUTEVIEWS

A collaborative router looking glass to share BGP views among network operators and researchers.

slide-3
SLIDE 3

ROUTEVIEWS

ROUTEVIEWS

A collaborative router looking glass to share BGP views among network operators and researchers. RouteViews was founded at the University

  • f Oregon’s Advanced Network Technology

Center (ANTC) in 1995. Data archives began in 1997 and amount to 22TBs (compressed) today.

slide-4
SLIDE 4

ROUTEVIEWS

ROUTEVIEWS

A collaborative router looking glass to share BGP views among network operators and researchers. RouteViews was founded at the University

  • f Oregon’s Advanced Network Technology

Center (ANTC) in 1995. Data archives began in 1997 and amount to 22TBs (compressed) today. The group is currently led by the network engineering team at the University of Oregon with assistance from the Network Startup Resource Center (NSRC) group.

slide-5
SLIDE 5

ROUTEVIEWS

A collaborative router looking glass to share BGP views among network operators and researchers. RouteViews was founded at the University

  • f Oregon’s Advanced Network Technology

Center (ANTC) in 1995. Data archives began in 1997 and amount to 22TBs (compressed) today. The group is currently led by the network engineering team at the University of Oregon with assistance from the Network Startup Resource Center (NSRC) group.

ROUTEVIEWS

NSRC supports the growth of global Internet infrastructure by providing engineering assistance, collaborative technical workshops, training, and

  • ther resources to university, research & education

networks worldwide. NSRC is partially funded by the IRNC program of the NSF and Google with

  • ther contributions from public and private
  • rganizations.

NSRC

slide-6
SLIDE 6

ROUTEVIEWS

The University of Oregon is a public research institution in Eugene, Oregon, USA founded in

  • 1876. UO is renowned for its research prowess

and commitment to teaching. Both NSRC and RouteViews are based at the UO.

ROUTEVIEWS

A collaborative router looking glass to share BGP views among network operators and researchers. RouteViews was founded at the University

  • f Oregon’s Advanced Network Technology

Center (ANTC) in 1995. Data archives began in 1997 and amount to 22TBs (compressed) today. The group is currently led by the network engineering team at the University of Oregon with assistance from the Network Startup Resource Center (NSRC) group. NSRC supports the growth of global Internet infrastructure by providing engineering assistance, collaborative technical workshops, training, and

  • ther resources to university, research & education

networks worldwide. NSRC is partially funded by the IRNC program of the NSF and Google with

  • ther contributions from public and private
  • rganizations.

NSRC UNIVERSITY OF OREGON

slide-7
SLIDE 7

ROUTEVIEWS

JOHN KEMP DAVE MEYER RANDY BUSH KIMBERLY (KC) CLAFFY LUCY LYNCH HANS KUHN JOEL JAEGGLI JOHN HEASLEY

SPECIAL THANKS

ROUTEVIEWS

slide-8
SLIDE 8

FOOTPRINT

ROUTEVIEWS

slide-9
SLIDE 9

ROUTEVIEWS

COLLECTOR LOCATIONS

 Atlanta (digital realty)  Chicago (equinx)  Chile  DC (eqix)  Eugene (Multi-hop)  Johannesburg (JINX, NAPAfrica)  London (LINX)  Miami (flix)  Nairobi (kixp)  Palo Alto (PAIX)  Perth (WAIX)  Portland (NWAX)  Sao Paulo (IX.br x4)  San Francisco (sfmix)  Singapore (Equinix SG)  Serbia (sox)  Sydney (equinix)  Tokyo (DIX-IE)  Cape Town

FOOTPRINT

slide-10
SLIDE 10

PEERING STATS

ROUTEVIEWS

slide-11
SLIDE 11

PEERING STATS

ROUTEVIEWS

314,486,083

TOTAL PREFIXES

slide-12
SLIDE 12

PEERING STATS

ROUTEVIEWS

314,486,083 833

TOTAL PREFIXES PEERING SESSIONS

slide-13
SLIDE 13

PEERING STATS

ROUTEVIEWS

314,486,083 833 239

TOTAL PREFIXES PEERING SESSIONS AUTONOMOUS SYSTEMS routeviews.org/peers/peering-status.html More peering information:

slide-14
SLIDE 14

COLLECTORS

ROUTEVIEWS

Commodity

  • 8-16 Cores
  • 32G-64G Ram
  • 400GB-1TB SSD
  • 1/10 GB eth

Vendor

  • ASR 1004

HARDWARE SOFTWARE

OpenSource

  • Linux/Centos and…
  • Quagga – bgpd
  • FRR – bgpd

Vendor

  • IOS XE
slide-15
SLIDE 15

COLLECTORS OPERATIONS

ROUTEVIEWS

Pros

  • If you can reach the

collector, you can peer Cons

  • Peerings are subject

to the routing anomalies that RouteViews seeks to

  • bserve and collect

MULTI-HOP INTERNET EXCHANGE

Pros

  • Better positioned to

address multi-hop issues

  • Geographic diversity
  • Peering diversity
slide-16
SLIDE 16

COLLECTOR DATA

ROUTEVIEWS

Multi-Threaded Routing Toolkit

  • https://tools.ietf.org/html/rfc6396
  • MRT provides a standard for parsing or dumping routing

information to a binary file.

  • RouteViews Dumps consist of BGP RIBs and UPDATES.
  • RIBs are dumped every 2 hours
  • UPDATEs are dumped every 15 minutes

MRT

slide-17
SLIDE 17

DATA ACCESS

ROUTEVIEWS

  • MRT files are bzipped and rsynced back to

http://archive.routeviews.org/ regularly

  • They can be accessed via, http, ftp and rsync.
slide-18
SLIDE 18

MRT TOOLS

ROUTEVIEWS

RIPE libBGPdump, UCLA BGP Parser, NTT BGPdump2, etc:

  • https://bitbucket.org/ripencc/bgpdump/wiki/Home
  • https://github.com/cawka/bgpparser
  • https://github.com/yasuhiro-ohara-ntt/bgpdump2
  • https://github.com/t2mune/mrtparse (Python)
  • https://github.com/rfc1036/zebra-dump-parser (Perl)
slide-19
SLIDE 19

COLLECTOR ACCESSIBILITY

ROUTEVIEWS

telnet://route-views*.routeviews.org

  • No username necessary.
  • Users are able to run show commands, e.g. show ip bgp x.x.x.x/x.
  • Why not SSH?!
  • RouteViews data is publicly available. We’ve got nothing to hide.
  • We use ssh for host management.
  • show ip route x.x.x.x next-hop is incorrect!
  • Remember, this is a collector. There’s no data-plane, thus no true

FIB.

GOTCHAS

slide-20
SLIDE 20

USE CASES

ROUTEVIEWS

  • BGP is the backbone of the Global Routing Infrastructure.
  • To ensure it's stability, it needs to be constantly monitored.
  • RouteViews provides:
  • Command-Line/ Looking Glass
  • Prefix Visibility, Verify Convergence, Path Stability
  • Comparing Local/Regional/Global Views
  • Troubleshooting Reachability

OPERATIONS

slide-21
SLIDE 21

USE CASES

ROUTEVIEWS

  • BGP anomalies and dynamics are critical as well.
  • RouteViews Provides:
  • Network Topology Monitoring
  • Route Leaks/Hi-Jacks (ex. Artemis, Cyclops)
  • Network Optimization
  • Growth, Aggregation, etc. In AS/V4/V6
  • Address Provenance
  • ~500 research publications have used RouteViews data
  • More info: http://www.routeviews.org/routeviews/index.php/papers/

RESEARCH

slide-22
SLIDE 22

ROUTEVIEWS

  • File-Based storage, MRT data format

BGP DATA DISTRIBUTION EVOLUTION

Generation Characteristics (current)

1st

slide-23
SLIDE 23

ROUTEVIEWS

  • File-Based storage, MRT data format
  • Asynchronous

BGP DATA DISTRIBUTION EVOLUTION

Generation Characteristics (current)

1st

slide-24
SLIDE 24

ROUTEVIEWS

  • File-Based storage, MRT data format
  • Asynchronous
  • Manual retrieval, sequencing, and consolidation

BGP DATA DISTRIBUTION EVOLUTION

Generation Characteristics (current)

1st

slide-25
SLIDE 25

ROUTEVIEWS

  • File-Based storage, MRT data format
  • Asynchronous
  • Manual retrieval, sequencing, and consolidation
  • No post-processing

BGP DATA DISTRIBUTION EVOLUTION

Generation Characteristics (current)

1st

slide-26
SLIDE 26

ROUTEVIEWS

  • File-Based storage, MRT data format
  • Asynchronous
  • Manual retrieval, sequencing, and consolidation
  • No post-processing
  • Centralized model

BGP DATA DISTRIBUTION EVOLUTION

Generation Characteristics (current)

1st

slide-27
SLIDE 27

ROUTEVIEWS

  • “Message-based” data distribution, per-message timestamps, with

meta-data

Generation Characteristics (future)

BGP DATA DISTRIBUTION EVOLUTION 2nd

slide-28
SLIDE 28

ROUTEVIEWS

  • “Message-based” data distribution, per-message timestamps, with

meta-data

  • Automated consolidating and sequencing

Generation Characteristics (future)

BGP DATA DISTRIBUTION EVOLUTION 2nd

slide-29
SLIDE 29

ROUTEVIEWS

  • “Message-based” data distribution, per-message timestamps, with

meta-data

  • Automated consolidating and sequencing
  • Database storage and access

Generation Characteristics (future)

BGP DATA DISTRIBUTION EVOLUTION 2nd

slide-30
SLIDE 30

ROUTEVIEWS

  • “Message-based” data distribution, per-message timestamps, with

meta-data

  • Automated consolidating and sequencing
  • Database storage and access
  • RESTful interfaces

Generation Characteristics (future)

BGP DATA DISTRIBUTION EVOLUTION 2nd

slide-31
SLIDE 31

ROUTEVIEWS

  • “Message-based” data distribution, per-message timestamps, with

meta-data

  • Automated consolidating and sequencing
  • Database storage and access
  • RESTful interfaces
  • Real-time streaming telemetry

Generation Characteristics (future)

BGP DATA DISTRIBUTION EVOLUTION 2nd

slide-32
SLIDE 32

ROUTEVIEWS

  • “Message-based” data distribution, per-message timestamps, with

meta-data

  • Automated consolidating and sequencing
  • Database storage and access
  • RESTful interfaces
  • Real-time streaming telemetry
  • Middle-layer abstraction, multi-client access (facilitates analysis and

services)

Generation Characteristics (future)

BGP DATA DISTRIBUTION EVOLUTION 2nd

slide-33
SLIDE 33

ROUTEVIEWS

  • “Message-based” data distribution, per-message timestamps, with

meta-data

  • Automated consolidating and sequencing
  • Database storage and access
  • RESTful interfaces
  • Real-time streaming telemetry
  • Middle-layer abstraction, multi-client access (facilitates analysis and

services)

  • RPKI validation

Generation Characteristics (future)

BGP DATA DISTRIBUTION EVOLUTION 2nd

slide-34
SLIDE 34

NEXT STEPS COMMUNICATION

ROUTEVIEWS

  • Better communications for those who are

interested.

  • Maintenance.
  • Outages.
  • Collector announcements.

User List

slide-35
SLIDE 35

NEXT STEPS GOVERNANCE

ROUTEVIEWS

  • Ensure RouteViews continues to meet the needs
  • f the community.
  • Comprised of research and industry members.

Steering Committee

slide-36
SLIDE 36

NEXT STEPS BMP & OpenBMP

ROUTEVIEWS

BGP Monitoring Protocol

  • https://tools.ietf.org/html/rfc7854
  • Available now – Cisco, Juniper, (FRR coming soon)
  • In addition to MRT attributes BMPs adds
  • Start, Stop, Peer Up, Peer Down
  • Collector Identification
  • Statistics

BMP

slide-37
SLIDE 37

NEXT STEPS BMP & OpenBMP

ROUTEVIEWS

  • BMP is the IETF standard for BGP monitoring
  • OpenBMPd is OpenSource (part of the Linux Foundation)
  • Consolidates peers/collectors
  • Splits collector, peer and update messages into separate

streams

  • Apache Kafka comprises the message bus for openbmp
  • Addresses producer/consumer problems
  • Proven to Scale
  • Mature client API
  • Clients in 16 different programming languages
  • Can be easily extended to meet future needs.
slide-38
SLIDE 38

OpenBMP ARHITECTURE

ROUTEVIEWS

https://github.com/OpenBMP/openbmp/blob/master/docs/images/openbmp-flow.png

slide-39
SLIDE 39

BMP TOOLS

ROUTEVIEWS

  • https://bgpstream.caida.org/

Languages:

  • https://cwiki.apache.org/confluence/display/KAFKA/Clients
slide-40
SLIDE 40

RESEACH OPPORTUNITIES

ROUTEVIEWS

By leveraging the 2nd generation characteristics of RouteViews BGP data distribution, new and novel approaches to BGP anomaly and dynamics analysis are possible.

Generation

2nd

slide-41
SLIDE 41

RESEACH OPPORTUNITIES

ROUTEVIEWS

  • Use RouteViews API data for ML supervised learning. Train models to

better detect:

  • Route leaking/hijacking
  • Infrastructure/peering outages
  • Internet censorship
  • Routing policy complexity
  • Validate ML models against live BMP streams

Generation

2nd

slide-42
SLIDE 42

THANK YOU

ROUTEVIEWS

Questions?