BGP Stream: An Analysis of One Year of BGP Incidents

04/03/2024

By Alejandro Acosta, R&D Coordinator at LACNIC

LACNIC presents the first webpage designed to show incidents and an analysis of Border Gateway Protocol (BGP) measurement data in Latin America and the Caribbean.

MAIN INCIDENTS. In addition to a summary of the information, the page shows three main types of events: possible network hijacks, BGP outages, and route leaks.

Possible hijacks refers to the illegitimate takeover of groups of IP addresses by corrupting Internet routing tables. This typically occurs when an Autonomous System announces a prefix that it does not originate.

Outages refers to the loss of visibility of network prefixes by a majority group of sensors.

Route leaks, as the name suggests, refers to the —potentially— unintentional announcement of a network prefix via BGP. For example, in a private peering traffic exchange, when one of the participants announces the peer’s prefix to the Internet. This case is the most difficult for algorithms to detect, so some of these incidents are not identified.

How is the data obtained?

This initiative uses Cisco BGP Stream, an automated process that selects the largest and most important incidents, providing information on the nature of the event and the ASNs involved.

Additional reading:

A Much-Needed BGP RFC: AS Path Prepending

The information is openly published, as LACNIC believes that it is important for engineers, network administrators, and organizations to gain insights into the most common incidents in the region and raise awareness about the situation.

This allows quickly investigating events, the rapid development of complex prototypes and tools, as well as large-scale monitoring applications (e.g., detecting connectivity outages, attacks, or BGP hijacks).

Using a system developed by LACNIC’s R&D department, raw data is collected, plotted, identified, cleaned, stored in a database, and later used to produce statistics and graphs. This occurs automatically every 24 hours.

RESULTS. During the study period —February 2023 to February 2024— we found the results shown in the charts below, which compare BGP events worldwide vs BGP events in our region.

A comparison between the global chart and the chart specific to the LAC region shows a similar pattern in the order of the most common incidents, with outages being the most frequent type of incident, followed by possible hijacks, and finally prefix leaks. It should also be noted that outages represent a higher percentage of the total number of incidents in our region than at the global level.

An analysis of the results table showing worldwide BGP events vs BGP events in our region reveals the following:

TOP 5 countries in our region with the highest number of BGP outages

Outages
CC	Events
BR	781
AR	99
HT	24
MX	22
CL	17

TOP 5 countries in our region with the highest number of possible Hijacks

Expected CC	Detected CC	Events
BR	BR	67
BR	none	35
PY	BR	24
BR	US	22
BR	CN	9

TOP 3 countries in our region with the highest number of route leaks

Origin CC	Leaker CC	Events
VE	VE	7
MX	MX	5
CL	PA	2

Impact

In this first year of operation, LACNIC has observed a reduction in BGP incidents. Several reasons for this have been identified, including a) the deployment and adoption of Resource Certification (RPKI), b) LACNIC’s Internet Routing Registry (IRR), and the adoption of RFC 9234 (Route Leak Prevention and Detection Using Roles in UPDATE and OPEN Messages).

The adoption of these tools is being driven by better operator practices and ISOC’s promotion of MANRS.

Conclusions

Possible hijacks, outages, and route leaks are the most common types of BGP incidents. During the initial year of data collection, a decrease in the number of cases was observed. However, it is expected that they will not disappear entirely in the near future. Implementing robust redundancy and resiliency measures in networks is crucial, as is the early detection and prevention of possible hijacks to ensure the integrity and reliability of Internet routes.

At LACNIC, our goal is to raise awareness and encourage ISPs and organizations to be prepared to handle these incidents efficiently when they occur.

References

https://stats.labs.lacnic.net/BGP/bgpstream-lac-region.html

https://stats.labs.lacnic.net/BGP/bgpstream.html

https://bgpstream.crosswork.cisco.com/ 

IPv6 Is Not Encrypted but Respects Privacy

To know more About pink line metro route Visit

Solved: nping - libnsock mksock_bind_addr(): Bind to 2001:db8:1::1:0 failed (IOD #1): Cannot assign requested address (99)

 Problem:

  When using nping (which comes with nmap) we receive the message:

libnsock mksock_bind_addr(): Bind to 2001:db8:1::1:0 failed (IOD #1): Cannot assign requested address (99)

Current situation:

  The issue is that nping can not bind to the IP 2001:db8:1::1

Solution:

  There could be many different solutions, I'm only going to mention one.  What I did was to create a logical interface (tunnel type) with the IPv6 address I was needing (2001:db8:1::1). Here you have the commands.

ip tuntap add mode tun dev tun1

ip -6 addr add 2001:db8:1::1 dev tun1

ifconfig tun1 up

Finally, you could execute something like:

nping -6 -S 2001:db8:1::1 --tcp-connect -c 2 -p 53 <ipv6_dest> --source-mac 00:50:XX:XX:XX:35  --dest-mac 2c:XX:XX:XX:44:20

Hope this is useful

Python Script: Probably useless but functional IPv6 Network scanner

Below is the code of what is probably useless but a functional IPv6 host scanner written in Python using threading.

To perform a regular (brute force) network scans in an IPv6 Network is almost impossible it can take over 5.000 years to finish.

This project was purely academic and I just wanted to learn about threading in Python.

This software is not recommended for general usage.....

This  script  will call the OS to actually perform the ping

This software receives two parameters:
a) Prefix to scan in the format 2001:db8::/64 (subnet, not host)
b) Number of simultaneous processes it can run (MAXPINGS)

One more time it was purely academic stuff but hopefully it can make your day

Finally, AFAIK nmap does not yet support IPv6 network scan.

The code written in python3:

--- cut here ---

#!/usr/bin/python3

import threading
import sys
import ipaddress
import subprocess
import time

CURRENTPINGS=0 # Number of simultaneous ping at a time

def DOPING6(IPv6ADDRESS):
  global MAXPINGS, CURRENTPINGS
  CURRENTPINGS+=1
  CMD="ping6 -c 3 "+str(IPv6ADDRESS) + " 2> /dev/null > /dev/null"
  return_code = subprocess.call(CMD, shell=True)
  if return_code == 0:  #If ping was succesful
    print (IPv6ADDRESS," is alive")
 
  CURRENTPINGS-=1
 
def main():
  global MAXPINGS, CURRENTPINGS
  if len(sys.argv) != 3: #Validate how many parameters we are receiving
    print("  Not enough or too many parameter")
    print("  Usage: ./scanipv6.py IPv6Prefix/lenght MAXPINGS")
    print("  Example: ./scanipv6.py 2001:db8::/64 20")
    print("  Prefix lenght can be between 64-128")
    print("  MAXPINGS corresponds to how many pings will be running at the same time")
    exit()

  SUBNET,MASK=sys.argv[1].split("/")
  MAXPINGS=int(sys.argv[2])

  for addr in ipaddress.IPv6Network(sys.argv[1]):  #Let's loop for each address in the Block
    ping_thread=threading.Thread(target=DOPING6,args=(addr,))

    while CURRENTPINGS >= MAXPINGS: # With this while we make it possible to run max simultaneous pings
      time.sleep(1)  # Let's wait one second before proceeding
      #print ("Interrumping...., CURRENTPINGS > MAXPINGS") #Uncomment this line just for debugging

    ping_thread.start()

main()