analysis of country wide internet outages caused by
play

Analysis of Country-wide Internet Outages Caused by Censorship - PowerPoint PPT Presentation

CAIDA Workshop on BGP and Traceroute data August 22nd, 2011- San Diego (CA), USA Analysis of Country-wide Internet Outages Caused by Censorship Alberto Dainotti - alberto@unina.it University of Napoli Federico II These slides are based on


  1. CAIDA Workshop on BGP and Traceroute data August 22nd, 2011- San Diego (CA), USA Analysis of Country-wide Internet Outages Caused by Censorship Alberto Dainotti - alberto@unina.it University of Napoli “Federico II” These slides are based on the following paper to be presented at ACM IMC 2011: A. Dainotti, C. Squarcella, E. Aben, K. C. Claffy, M. Chiesa, M. Russo, A. Pescapé, “Analysis of Country-wide Internet Outages Caused by Censorship” w w w . cai da. or g

  2. THE EVENTS Internet Disruptions in North Africa • Egypt - Protests in the country start around January 25th, 2011 - The government orders service providers to “shutdown” the Internet - On January 27th, around 22:34 GMT , several sources report the withdrawal in the Internet’s global routing table of almost all routes to Egyptian networks - The disruption lasts 5.5 days • Libya - Protests in the country start around 17th February 2011 - The government controls most of the country’s communication infrastructure - Three different connectivity disruptions: February 18th (6.8 hrs), 19th (8.3 hrs), March 3rd (3.7 days) • Similar events in other countries but we did not analyze them COMICS Research Group University of Napoli “Federico II” - Italy

  3. SOME FACTS Prefixes, ASes, Filtering Egypt - 3165 IPv4 and 6 IPv6 prefixes are delegated to Egypt by AfriNIC - They are managed by 51 Autonomous Systems - Filtering type: BGP only - Filtering dynamic: synchronized; progressive Libya - 13 IPv4 prefixes, no IPv6 prefixes - 2 (+ 1) Autonomous Systems operate in the country - Filtering type: mix of BGP , packet filtering, satellite signal jamming - Filtering dynamic: testing different techniques; somehow synchronized COMICS Research Group University of Napoli “Federico II” - Italy

  4. WHAT WE DID Combined different measurement sources • BGP - BGP updates from route collectors of RIPE-NCC RIS and RouteViews - We combined information from both databases - Graphical Tools: REX , BGPlay , BGPviz • Active Traceroute Probing - Archipelago Measurement Infrastructure ( ARK ) - We underutilized it.. • Internet Background Radiation (IBR) - Traffic reaching the UCSD network telescope - Capable of revealing different kinds of blocking COMICS Research Group w w w . cai da. or g University of Napoli “Federico II” - Italy

  5. THE DATA Geolocation + announced prefixes • IP ranges associated with the country of interest - Delegations from Regional Internet Registries (RIR) - Commercial geolocation database Egypt Libya AfriNIC delegated IPs 5,762,816 299,008 MaxMind GeoLite IPs 5,710,240 307,225 • Gather prefixes to be monitored. For each IP range: - We look up the address space in the BGP database of announced prefixes, to find an exactly matching BGP prefix - We find all the more specific (strict subset, longer) prefixes of this prefix - If the two previous steps yielded no prefix, we retrieve the longest BGP prefix entirely containing the address space • Every time we refer to an AS we actually refer to the IPs of that AS that are associated to the country of interest COMICS Research Group University of Napoli “Federico II” - Italy

  6. BGP prefix reachability • We reconstruct prefixes losing and regaining reachability - we build the routing history of a collector’s peer for each collector - using both RIBs and UPDATES - we mark a prefix as disappeared if it is withdrawn in each routing history Egyptian disconnection and reconnection [NOTE: IPv6 routes stayed up!] 3500 3500 number of re-announced IPv4 prefixes 3000 3000 number of visible IPv4 prefixes 2500 2500 2000 2000 1500 1500 1000 1000 500 500 0 0 20:00 20:30 21:00 21:30 22:00 22:30 23:00 09:00 09:30 10:00 10:30 11:00 11:30 12:00 COMICS Research Group University of Napoli “Federico II” - Italy

  7. BGP per-AS analysis • A detailed analysis shows there is synchronization among ASes EgAS1 EgAS2 EgAS3 EgAS4 EgAS5 EgStateAS 1000 number of re-announced IPv4 prefixes 800 600 400 200 0 09:00 09:30 10:00 10:30 11:00 11:30 12:00 COMICS Research Group University of Napoli “Federico II” - Italy

  8. ROUTE CHANGES BGPlay • The massive disconnection caused some path changes too COMICS Research Group University of Napoli “Federico II” - Italy

  9. UCSD TELESCOPE when malware helps.. • Unsolicited traffic - e.g. scanning from conficker-infected hosts - from the observed country and reaching a (mostly) unused /8 network at UCSD Egypt Libya 140 120 100 packets per second 80 60 40 20 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 2 2 2 2 - - - - - - - - - 2 2 2 3 3 0 0 0 0 7 8 9 0 1 1 2 3 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 : : : : : : : : : 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 COMICS Research Group University of Napoli “Federico II” - Italy

  10. UCSD TELESCOPE need to dissect traffic • We classified traffic to the telescope in - Conficker-like - Backscatter (e.g. SYN-ACKs to randomly spoofed SYNs of DoS attacks) - Other Egypt: telescope traffic 80 70 60 packets per second 50 40 30 20 10 0 01-27 00:00 01-28 00:00 01-29 00:00 01-30 00:00 01-31 00:00 02-01 00:00 02-02 00:00 02-03 00:00 02-04 00:00 COMICS Research Group University of Napoli “Federico II” - Italy other conficker-like backscatter

  11. TELESCOPE VS BGP Co nsistency • The sample case of EgAS7 shows the consistency between telescope traffic and BGP measurements Egypt: disconnection of EgAS7 0.7 100 0.6 Number of IPv4 prefixes in BGP 80 0.5 packets per second 60 0.4 0.3 40 0.2 20 0.1 0 0 01-27 00:00 01-28 00:00 01-29 00:00 01-30 00:00 01-31 00:00 02-01 00:00 02-02 00:00 02-03 00:00 02-04 00:00 COMICS Research Group packet rate of unsolicited traffic University of Napoli “Federico II” - Italy visibility of BGP prefixes

  12. TELESCOPE VS BGP Co mplementarity Libya 8 • Contrasting telescope traffic with 7 BGP measurements revealed a mix of 6 packets per second 5 blocking techniques that was not 4 publicized by others 3 2 • The second Libyan outage involved 1 overlapping of BGP withdrawals 0 02-18 12:00 02-19 00:00 02-19 12:00 02-20 00:00 02-20 12:00 02-21 00:00 and packet filtering 14 12 number of visible prefixes 10 8 6 4 2 COMICS Research Group 0 02-18 12:00 02-19 00:00 02-19 12:00 02-20 00:00 02-20 12:00 02-21 00:00 University of Napoli “Federico II” - Italy

  13. TELESCOPE VS BGP Co nfusion? • BGP-unreachability doesn’t, in general, prevent outbound traffic - We found networks that were BGP-unreachable sending traffic to the telescope - and networks BGP-reachable that were not - Topology analysis may help to better understand this behavior Telescope traffic from two Egyptian ASes 90 80 70 packets per second 60 50 40 30 20 10 0 01-27 00:00 01-28 00:00 01-29 00:00 01-30 00:00 01-31 00:00 02-01 00:00 02-02 00:00 02-03 00:00 02-04 00:00 COMICS Research Group EgAS4 EgStateAS University of Napoli “Federico II” - Italy

  14. ARK active measurements • ARK active measurements are consistent with other sources - limitation due to frequency of probes and because they target random addresses - the first two Libyan outages are not visible - we used them only to test reachability , not to analyze topology Egypt Libya ! ! 5% ! ! ! ! ! ! ! ! ! Ark traceroute to Egypt terminating in Egypt Ark traceroute to Libya terminating in Libya ! ! ! 15% ! ! ! ! 4% ! ! ! ! ! ! ! ! ! 3% 10% ! ! ! 2% ! 5% ! ! 1% ! ! ! ! Feb Mar Feb Mar ! ! ! ! ! ! ! ! ! ! ! ! 12 14 16 18 20 22 24 26 28 2 4 6 8 10 12 14 Jan Jan Feb Feb 21 22 23 24 25 26 27 28 29 30 2 3 4 5 6 COMICS Research Group University of Napoli “Federico II” - Italy

  15. ARK confirming telescope’s findings • Third Libyan outage: while BGP reachability was up, most of Libya was disconnected - ARK measurements confirmed the finding from the telescope, plus identified some reachable hosts, suggesting the use of packet filtering by the censors Libya: ARK (left) , Telescope (right) 5% ! 8 ! Ark traceroute to Libya terminating in Libya ! ! ! ! 7 ! 4% ! ! ! ! ! 6 ! ! ! packets per second ! 3% 5 ! ! 4 ! 2% 3 ! 2 1% 1 ! ! ! ! Feb Mar Feb Mar ! ! ! ! ! ! 0 ! ! 03-01 00:00 03-02 00:00 03-03 00:00 03-04 00:00 03-05 00:00 03-06 00:00 03-07 00:00 03-08 00:00 12 14 16 18 20 22 24 26 28 2 4 6 8 10 12 14 COMICS Research Group University of Napoli “Federico II” - Italy

  16. SATELLITE CONNECTIVITY probable signal jamming • Third Libyan outage - a Libyan IPv4 prefix managed by SatAS1 was BGP-reachable - a small amount of traffic from that prefix reaches the telescope Libya: Telescope traffic from national operator and satellite-based ISP 5 4 packets per second 3 2 1 0 0 0 0 0 2 2 3 3 - - - - 1 2 0 1 9 6 4 1 0 0 0 0 0 0 0 0 : : : : 0 0 0 0 0 0 0 0 COMICS Research Group University of Napoli “Federico II” - Italy LyStateAS SatAS1

Recommend


More recommend