Is the security of BGP using a pre-shared key dependent on the security of MD5?

Question

The german wikipedia article to BGP states (translated):

In the basic configuration, a BGP-Router is susceptible to spoofig attacks, through which the attacker can manipulate the routers. By using an authentiation between the BGP peers with a individual password (based on MD5 hashes), the data exchange between BGP routers can be secured. This makes it harder to use spoofic attacs, but it is especially dependent on the security of MD5, which is not seen as secure by crypto experts anymore. (https://de.wikipedia.org/wiki/Border_Gateway_Protocol)

How exactly is the security dependent on the security of MD5? Like the password/pre-shared key would need to be stored on both routers (one sends it - the other verifys it). So if the hash is actually stolen, it could be used without any need for decryption, or am I wrong with that? So how would anyone benefit from using a stronger hashing algorithm (or especially how does the security flaws of MD5 make the pre-shared key less secure)?

Answer 1

Yes, BGP peers that use MD5 based password authentication are subject to MD5 encryption exploit. If the MD5 secret is not sufficiently complex or is already present in MD5 hash lookup tables it may be potentially exploited. But the password alone is not the only factor.

MD5 authentication for BGP peers also includes the IP address of the peer (which can also potentially be spoofed), as well as a hash of the BGP packet itself. So there are 3 factors that need to be intercepted and spoofed to effectively exploit an MD5 peer setup. It is not very trivial under a good setup, especially if the password is complex and not commonly used.

Additionally BGP peering sessions will/should make use of filtering via access lists or similar methods to define what route data is acceptable for the peer relationship. This can make even a successful exploit relatively useless beyond possibly a denial of service or other annoyance. A lot of work for relatively little return value. Usually BGP exploits involve peers that do not have authentication or announcement verification/filtering. Just implementing the basic security features can deter a lot of abusers.

There are additional BGP security features that depend on the platform in use and the security posture of the organization in question. You can see a lot of the information in this Ciscopress document:

https://www.ciscopress.com/articles/article.asp?p=1312796&seqNum=3