I agree phishing bgp feed would disrupt the ip address to all ISP's that listened to the bgp server involved. I was addressing a specific issue with listening to such a server and that is the loss of control issue. Sorry if that wasn't clear. So would ISP's block an phishing site if it was proven to be a phishing site and reported by their customers? Donald.Smith@qwest.com GCIA pgpFingerPrint:9CE4 227B B9B3 601F B500 D076 43F1 0767 AF00 EDCC Brian Kernighan jokingly named it the Uniplexed Information and Computing System (UNICS) as a pun on MULTICS.

-----Original Message----- From: Stephen J. Wilcox [mailto:steve@telecomplete.co.uk] Sent: Monday, June 28, 2004 2:58 PM To: Smith, Donald Cc: Scott Call; nanog@nanog.org Subject: RE: BGP list of phishing sites?

Hi Donald, the bogon feed is not supposed to be causing any form of disruption, the purpose of a phishing bgp feed is to disrupt the IP address.. thats a major difference and has a lot of implications.

Steve

On Mon, 28 Jun 2004, Smith, Donald wrote:

...

Some are making this too hard. Of the lists I know of they only blackhole KNOWN active attacking or victim sites (bot controllers, know malware download locations etc) not porn/kiddie porn/pr/choose-who-you-hate-sites ... clients (infected pc's) are usually not included but could make it on the list given enough attacks. It does mean giving up some control of your network which may not be acceptable to some ISP's. Its not much different then listening to an automated bogon feed.

Donald.Smith@qwest.com GCIA pgpFingerPrint:9CE4 227B B9B3 601F B500 D076 43F1 0767 AF00 EDCC Brian Kernighan jokingly named it the Uniplexed Information and Computing System (UNICS) as a pun on MULTICS.

...

-----Original Message----- From: owner-nanog@merit.edu [mailto:owner-nanog@merit.edu] On Behalf Of Stephen J. Wilcox Sent: Monday, June 28, 2004 11:56 AM To: Scott Call Cc: nanog@nanog.org Subject: Re: BGP list of phishing sites?

On Sun, 27 Jun 2004, Scott Call wrote:

...

On the the things the article mentioned is that ISP/NSPs are shutting off access to the web site in russia where the malware is being downloaded from.

Now we've done this in the past when a known target of a DDOS was upcoming or a known website hosted part of a malware package, and it is fairly effective in stopping the problems.

So what I was curious about is would there be interest in a BGP feed (like the DNSBLs used to be) to null route known malicious sites like that?

Obviously, both operational guidelines, and trust of the operator would have to be established, but I was thinking it might be useful for a few purposes:

1> IP addresses of well known sources of malicious code (like in 1> the example above) 2> DDOS mitigation (ISP/NSP can request a null route of a prefix which will save the "Internet at large" as well as the NSP from the traffic flood 3> etc

Since the purpose of this list would be to identify and mitigate large scale threats, things like spammers, etc would be outside of it's charter.

If anyone things this is a good (or bad) idea, please let me know. Obviously it's not fully cooked yet, but I wanted to throw it out there.

Personally - bad.

So what do you want to include in this list.. phishing? But why not add bot C&C, bot clients, spam sources, child porn, warez sites. Or if you live in a censored region add foreign political sites, any porn, or other messages deemed bad.

Who maintains the feed, who checks the sites before adding them, who checks them before removing them.

What if the URL is a subdir of a major website such as aol.com or ebay.com or angelfire.com ... what if the URL is a subdir of a minor site, such as yours or mine?

What if there is some other dispute over a null'ed IP, suppose they win, can they be compensated?

Does this mean the banks and folks dont have to continue to remove these threats now if the ISP does it? Does it mean the bank can sue you if you fail to do it?

What if you leak the feed at your borders, I may not want to take this from you and now I'm accidentally null routing it to you. Should you leak this to downstream ASNs? Should you insist your Tier1 provides it and leaks it to you?.. just you or all customers?

What if someone mistypes an IP and accidentally nulls something real bad(TM)? What if someone compromises the feeder and injects prefixes maliciously?

What about when the phishers adapt and start changing DNS to point to different IPs quickly, will the system react quicker? Does that mean you apply less checks in order to get the null route out quicker? Is it just /32s or does it need to be larger prefixes in the future? Are there other ways conceivable to beat such a system if it became widespread (compare to spammer tactics)

What if this list gets to be large? Do we want huge amounts of /32s in our internal routing tables?

What if the feeder becomes a focus of attacks by those wishing to carry out phishing or other illegal activities? This has certainly become a hazard with spam RBLs.

Any other thoughts?

Steve