http://www.computerworld.com/securitytopics/security/story/0,10801,74385,00.html By VINCE TUESDAY SEPTEMBER 23, 2002 My security team was recently asked to help reduce costs by consolidating after-hours security and IT support services. We had been charging a nightly fee for round-the-clock on-call support, but the company reasons that it's cheaper to consolidate all first-line support to the on-call IT team that supports our applications. I trained the on-call team, covering the most common problems and what to do if there's a situation they can't handle. My security team now offers second-line support. The alerting system is well tuned, and we don't get many after-hours alarms, so I doubted we'd be called often. I was wrong. At 3 a.m. on the first night the IT team took over, I received a call from a rather worried on-call guy who had been paged with an "ISS" alert. He didn't know that ISS just stands for Internet Security Systems Inc., the Atlanta-based vendor of our intrusion-detection software. One of the many things we can detect is probes sent from ISS's Internet Scanner software. The scanner lets administrators check their networks for vulnerabilities, but attackers can also misuse it to map our networks and identify weaknesses. ISS tries to prevent this by using a complicated licensing process that limits the IP addresses each tool can attack. It also sends some special packets at the beginning of each scan, including the license key, the user name, and the host and domain of the scanning machine. That way, if someone uses the tool to scan a network they don't own, the product will announce who they are. We monitor for these packets in case somebody finds a way, using network address translation perhaps, to trick the scanner into thinking it's probing a local machine when in fact it's scanning us. More worrisome is that, as with other digital rights management systems, hackers claim to have broken ISS's license key system. In fact, key-generation software can be found on the Web to make keys for any network. The fake license keys these tools generate typically have an ID of 1234. So even if the special packets contain the hacked ID, you have very little to go on. We could also expect the attacker's IP address to be faked. The normal response to an ISS alert, we told the new support team, is to trace down the source of the attack via the America Registry for Internet Numbers (ARIN) Web site at www.arin.net and notify the attacker's Internet service provider. We even have standard forms for those submissions. We don't really expect the ISP to do anything, but at least we try. Internal Attack But the detail that worried the front-line support chap, and that made me snap awake at that awful hour, was the source of the attack: It came from within our own network. Maybe someone we'd hired was a bit of a hacker. Or maybe the system had it wrong and the attacker was actually the target. I asked the support technician for the target address of the probes. It was the IP address 11.1.1.3, which seemed rather odd. The address range 10.x.x.x is reserved so companies can use it internally, as we do. So perhaps this was a typo? Who was 11.x.x.x? After a quick check of ARIN, my blood ran cold. The results read: DOD Intel Information Systems (NET-DODIIS) Defense Intelligence Agency Washington, D.C. We had detected an attack against the DIA, the heart of the U.S. intelligence services, that came from our own network - and I doubted that we were the only people to spot this. No doubt somewhere in Washington someone was also being woken to respond. Whoops. We had to work out what was going on before men in trench coats and dark glasses arrived. I took control of the call and began searching for the internal machine. I traced the machine to one of our Unix server clusters. It seemed to be the one running our enterprise monitoring system (EMS). That didn't make sense - ISS stopped making Unix versions of Internet Scanner a long time ago. If a hacker could get hold of a license generator, why would he make keys for an old version? The Unix version of Internet Scanner was five years old, so the problems it might look for would have been fixed by now, making it useless. The Mix-up Then a few pieces fell into place. The EMS pings every interface on every router we have to make sure each is responding correctly, and the ISS special packets use the same protocol as ping. Perhaps there had been some kind of mix-up? Could the EMS, by chance, have sent an ISS alarm packet? I woke the network team and got them to check the configuration. Aha! We were monitoring the 11.1.1.3 address. It seems someone had mistyped what should have been a 10.x.x.x address. So our EMS was accidentally trying to manage the DIA's network devices. But was it also the unwitting host of a hacked version of Internet Scanner? We could find no evidence of any such tools on the machine. It seemed much more likely that the "attack" was just an odd packet. But with no record of the packet from our intrusion-detection system, we faced a choice: We could either ask the DIA if they had a copy, or we could keep our heads down. We're keeping our heads down. I've updated the firewall to block any attempts for the EMS to talk outward, and hopefully that's the end of it. That is, unless the feds come knocking. What do you think? This week's journal is written by a real security manager, "Vince Tuesday," whose name and employer have been disguised for obvious reasons. Contact him at vince.tuesdayat_private, or join the discussion in our forum: - ISN is currently hosted by Attrition.org To unsubscribe email majordomoat_private with 'unsubscribe isn' in the BODY of the mail.
This archive was generated by hypermail 2b30 : Fri Sep 27 2002 - 02:52:50 PDT