Low TTL Values

Any IP router is supposed to decrement the TTL value in the IP header by one9 every time an IP packet is forwarded to another system. If packets appear within your local subnet with a TTL value of one, then someone is most likely using the traceroute program (or a variant such as tcptraceroute) against an IP address that either exists in the local subnet or is in a subnet that is routed through the local subnet. Usually this is simply someone troubleshooting a network connectivity problem, but it can also be an instance of someone performing reconnaissance against your network in order to map out hops to a potential target.

NOTE Packets destined for multicast addresses (all addresses within the range 224.0.0.0

through 239.255.255.255, as defined by RFC 1112) commonly have TTL values set to one. So if the destination address is a multicast address, it is likely that such traffic is not associated with network mapping efforts with traceroute and is just legitimate multicast traffic.

A UDP packet produced by traceroute is logged as follows by iptables (note the TTL in bold):

Jul 24 01:10:55 iptablesfw kernel: DROP IN=eth0 OUT=

MAC=00:13:d3:38:b6:e4:00:13:46:c2:60:44:08:00 SRC=144.202.X.X DST=71.157.X.X LEN=40 TOS=0x00 PREC=0x00 TTL=1 ID=44081 PROTO=UDP SPT=54522 DPT=33438 LEN=20

8 Taking a host-centric view of intrusion detection is known as target-based intrusion detection, which allows an IDS to factor in implementation details of target systems; more on this in Chapter 8.

9 It is possible for a router to decrement the TTL value by two or more if the number of seconds the router holds onto the packet before forwarding it is greater than one second. RFC 791 states that a router must decrement the TTL by at least one.

CONCEALING AN ATTACK WITH FRAGMENTS

AND TARGETED TTLS

Routing path information is useful for concealing network attacks with fragment reassembly tricks. For example, suppose that an attacker sees that a router exists in front of a host (as determined with traceroute), and that the attacker also suspects that an IDS is watching the subnet that is in front of the host subnet. If this is the case, the host can be targeted with an attack that is fragmented over three IP packets (let's call them f1, f2, and f3), but in such a way that the attack is not detected by the IDS. The attacker can accomplish this by creating a duplicate of the second fragment (f2), replacing its payload with dummy data, and reducing its TTL to an initial value that is just large enough to get the packet to the router with a TTL of one. Let's call this packet f2'. Next, the attacker sends the first fragment (f1), followed by this new fragment (f2'), followed by f3, and finally, the original f2 fragment. Thus, the IDS (which is in front of the router) sees all four fragments, but f3 completes the set of fragments and hence the IDS reassembles them as f1 + f2' + f3.

Recall that f2' contains dummy data, so these three fragments together do not look like an attack to the IDS. Meanwhile, f2' hits the router and gets dropped because its TTL value is decremented to zero before it is forwarded, so the target IP address never sees f2'. However, the host has seen fragments f1 and f3, but it can't reassemble them to anything meaningful without the original f2, so it waits for it.

When f2 finally arrives (remember that the attacker sent it last), the target host is hit with the real attack after the host finally reassembles all three fragments. This technique was first proposed in "Bro: A System for Detecting Network Intruders in Real-Time" by Vern Paxson (see http://www.icir.org/vern/papers/bro-CN99.html); it provides a clever way to utilize the network layer to hide attacks from network intrusion detection systems.

NOTE Another suspicious TTL value for any packet on the local subnet is a TTL of zero.

Such a packet can only exist if there is either a severely buggy router that forwarded the packet into the subnet or the packet originated from a system on the same subnet.

Was this article helpful?

0 0

Post a comment