INTERNET HOLES - ELIMINATING IP ADDRESS FORGERY
COPYRIGHT (C), 1996, MANAGEMENT ANALYTICS - ALL RIGHTS RESERVED
_________________________________________________________________
Series Introduction
The Internet is now the world's most popular network and it is full of
potential vulnerabilities. In this series of articles, we explore the
vulnerabilities of the Internet and what you can do to mitigate them.
An Introduction IP Address Forgery
The Internet Protocol (IP) (RFC791) provides for two and only two
functions. It defines a datagram that can be routed through the
Internet, and it provides a means for fragmenting datagrams into
packets and reassembling packets into the original datagrams. To quote
from RFC791:
The internet protocol is specifically limited in scope to provide the
functions necessary to deliver a package of bits (an internet
datagram) from a source to a destination over an interconnected
system of networks. There are no mechanisms to augment end-to-end
data reliability, flow control, sequencing, or other services
commonly found in host-to-host protocols. The internet protocol
can capitalize on the services of its supporting networks to
provide various types and qualities of service.
Here's a description of an IP datagram, also from RFC791:
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|Version| IHL |Type of Service| Total Length |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Identification |Flags| Fragment Offset |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Time to Live | Protocol | Header Checksum |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Source Address |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Destination Address |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Options | Padding |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| data |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| data |
\ \
\ \
| data |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| data |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Description of an IP Datagram
Note that the 4th line of the description calls for the Source Address
of the datagram. In the simplest form of IP address forgery, the
forger only needs to create a packet containing a false Source Address
and insert it into the Internet by writing it into the output device
used to send information to the rest of the Internet. For the
non-expert forger, there is a tool called iptest which is part of the
free and publicly available ipfilter security package that
automatically forges packets for the purpose of testing configurations
or routers and other IP security setups.
The infrastructure of the Internet consists primarily of a set of
gateway computers and packet routers. These systems have multiple
hardware interfaces. They maintain routing tables to let them decide
which output interface to send a packet out on based on the input
interface that it came in on and the destination IP address specified
in the packet. When a forged packet arrives at an infrastructure
element, that element will faithfully route the packet toward the
destination address, exactly as it would a legitimate packet.
How Can IP Address Forgery Be Used
At its root, IP address forgery is a method of deception, and thus it
can be used in much the same way as other forms of deception.
Dunnigan95 More specifically, and using Dunnigan and Nofi's
classification scheme, here are some quick ideas about how IP address
forgery might be used:
* Concealment: IP address forgery is commonly used to conceal the
identity of an attacker, especially when denial of services is the
goal of the attack.
* Camouflage: IP address forgery is used to make one site appear to
be another as a way to convince the victim, for example, that an
attack is from a University, when in fact it is from a competitor.
* False and Planted Information: IP address forgery can be used to
create the impression that a particular site is acting maliciously
in order to create friction or lead a defender to falsely accuse
an innocent third party.
* Reuses: IP address forgery can be used to support another activity
designed to gain the confidence of the defender. For example, a
salesperson for information security products could create IP
address forgeries in order to convince a client of the need for
their services.
* Displays: IP address forgery has been used in order to lead
defenders to believe that many sites are participating in an
attack when in fact only a small number of individuals are
responsible.
* Demonstrations: IP address forgery has been used to demonstrate a
potential for untraceable attacks as a way to convince defenders
not to try to catch attackers.
* Feints: IP address forgery can be used to try to fool an enemy
into believing that an attack is coming from outside or from a
particular direction, when the real attack is very different. This
is a way to misdirect the enemy into spending limited resources in
the wrong way.
* Lies: IP address forgery has been used to create a more convincing
lie that somebody known to the defender is communicating with them
about a particular matter.
* Insight: IP address forgery can be used to gain insight into how
an opponent reacts and as a sort of probe to determine what sorts
of responses are likely to arise.
Another way to view this issue is in terms of the net effect on
information in information systems. Here is another way of viewing
this issue with an example from each category.
* Corruption of Information: IP addresses are often used as the
basis for Internet control decisions. For example, DNS updates are
often designated as coming only from specific other servers. With
IP address forgery, the entire DNS system could be corrupted,
causing services to be rerouted through enemy servers.
* Denial of Services: The Internet is basically a fragile network
that depends on the proper behavior and good will of the
participants for its proper operation. Without wide-ranging
changes to the way the Internet works, denial of services is
almost impossible to prevent. For example, the same DNS attack
could be used to cause widespread denial of services, or perhaps
even to create loops in the packet delivery mechanisms of the
Internet backbone.
* Leakage of Information: Forged IP addresses can be used to cause a
host to take orders for the delivery of information to enemy sites
by forging authorization as if it were from a legitimate
authorizing site.
* Misplaced Liability: Forged IP addresses could be used, as
described above under False and Planted Information, to cause
defenders to assert claims against innocent bystanders and to lay
blame at the wrong feet.
These are only some of the examples of what forged IP addresses can
do. Without a lot of effort, many other examples can be created.
What Can We Do About It?
As individuals, there is little we can do to eliminate all IP address
forgery, but as a community, we can be very effective. Here's how.
Instead of having all infrastructure elements route all packets, each
infrastructure element could, and should, enforce a simple rule. They
should only route packets from sources that could legitimately come
from the interface the packet arrives on.
This may sound complicated, but it really isn't. In fact, the
technology to do this is already in place, and always has been.
Virtually every router and gateway in existence today allows for the
filtering of packets based on their input interface and IP source and
destination address. This is a necessary component of their operation
and is the basis for the way they route all packets.
The only change that has to be made is for these routers and gateways
to enforce the network structure that is legitimately in place. Or in
other words, the routers and gateways should refuse to route
ridiculous packets. Here are some of the simpler examples of known bad
packets:
* The IP address 127.0.0.1 is ONLY used for internal routing of
packets from a host to itself. There is no legitimate IP datagram
that should pass through a router or gateway with this as the
source address. In fact, routing these packets is dangerous
because they may be used to forge packets from the localhost which
often has special privileges. A recent attack that causes denial
of services involves sending a packet to a host's echo port with
127.0.0.1 as its source address and the echo port as it's source
port. The echo port causes whatever packet it is sent to be
returned to its source. Since the source address is the same port
on the same host, this packet creates an infinite loop which, in
many cases, disables the computer.
* The IP address 0.0.0.0 is not legitimate - full stop. In fact,
there's really no legitimate IP address that should traverse
gateways containing a 0 for one of the address elements.
Unfortunately, many routers use the '.0.' convention in their
filtering tables to indicate any address from 0 to 255 (the whole
range), so blocking these packets may be non-trivial in some
infrastructure elements.
* The IP specification includes provisions for private subnetworks
that are designated for internal use only. There is no legitimate
reason to route packets from these addresses anywhere in the
general Internet infrastructure. (RFC1597) These address ranges
include 10.*.*.*, 172.16-32.*.*, and 192.168.*.* (where *
indicates any value from 0 through 255). No packets should be
routed through the Internet with these addresses as either their
source or their destination.
The next step in eliminating IP address forgery is for the routers and
gateways at each type of infrastructure element to enforce standards
on each interface. Generally, the Internet is broken up into Backbone
providers that provide wide area packet transport services, Private
Networks which are owned and operated by companies, institutions,
government agencies, and other parties for their own purposes, and
Internet Service Providers (ISPs) that provide connections between
the backbone elements and private networks (sometimes including other
ISPs). These roles can be blurred at times, but they are adequate for
our purposes.
* Private Networks: Each private network should;
+ 1) prevent all of the known-bad packets from crossing into or
out of the organization,
+ 2) prevent packets with internal source addresses from
passing inward,
+ 3) prevent packets with external source addresses from
passing outward,
+ 4) prevent packets with external destination addresses from
passing inward, and
+ 5) prevent packets with internal destination addresses from
passing outward.
* ISPs: Each ISP should;
+ 1) prevent all of the known-bad packets from crossing into or
out of their infrastructure,
+ 2) prevent any packet inbound from any of their clients with
a source address not from that client's assigned address
range from passing from the client network,
+ 3) prevent any packets with a destination address not in
their client's address range from passing to the client
network,
+ 4) prevent any packet not from this ISP's legitimate address
range from entering the backbone, and
+ 5) prevent any packets originating from the backbone and not
destined for one of their legitimate IP addresses from
entering their network.
Two additional rules will assist the ISP's clients;
+ 6) prevent inbound traffic from the client with the client's
address as a destination, and
+ 7) prevent outbound traffic to the client with the client's
address claimed to be the source.
* Backbone Networks: Each backbone provider should;
+ 1) prevent all of the known-bad packets from crossing into or
out of their infrastructure,
+ 2) prevent packets originating from any ISP with source
addresses not in that ISP's range of legitimate source
addresses from entering the backbone,
+ 3) prevent any packets not destined for an ISP's address
range from entering that ISP,
+ 4) prevent any packets from any other backbone provider that
could not be properly routed through that provider from
entering their backbone, and
+ 5) prevent any packets from going to any other backbone
provider unless they could legitimately be routed through
that provider to reach their destination.
For backbones, this requires some effort, however the high volume of
information they carry certainly justifies a little effort for
protection.
Some Examples
As an aide to the less technically inclined, the following examples
provide some real world implementation details.
This set of rules applies to a private network (in this case, the
all.net class C network 204.7.229.*) and are written in the format of
the Morningstar PPP (point to point protocol) Filter file:
# Rule 1 for private networks
# prevent known-bad address ranges from entering (or leaving)
!172.16-32.0.0 # private network segment
!192.168.0.0 # private network segment
!10.0.0.0 # private network segment
!127.0.0.0 # localhost network
# Rule 2 for private networks
# prevent internal source address packets from passing inward
!recv/src/204.7.229.0 # prevent inbound from our network
# Rule 5 for private networks
# prevent internal destination addresses from passing outward
# Note that rule 5 is placed here because the filters are order dependent
!send/dst/204.7.229.0 # prevent our destinations from passing out
# Rule 3 for private networks
# prevent external source address packets from passing outward
send/src/204.7.229.0 # allow legitimate outbound sources
!send/src/0.0.0.0 # prevent illegitimate outbound sources
# Rule 4 for private networks
# prevent external destinations from passing inward
recv/dst/204.7.229.0 # allow legitimate inbound destinations
!recv/dst/0.0.0.0 # prevent illegitimate inbound destinations
The next set of rules applies to an ISP. In this case, we assume that
the ISP has control over three class B networks that it uses to sell
services to its clients. The class B networks used in this example
have IP addresses of 123.7.*.*, 231.6.*.*, and 201.96.*.*. In this
case, we have three different parts of the example:
This is the router connecting the ISP to the backbone, presented in
the format of a Cisco router with interface 0 connected to the
backbone and interface 1 connected to the ISP's internal network. It
implements rules 1, 4, and 5 for the ISP.
# Rule 1 for an ISP
# prevent all of the known-bad address ranges
# this should be done on all in and out connections
# on all interfaces in all access control lists
All interfaces in and out
deny ip 172.16-32.0.0 # private network segment
deny ip 192.168.0.0 # private network segment
deny ip 10.0.0.0 # private network segment
deny ip 127.0.0.0 # localhost network
# Rule 2 for an ISP
# prevent inbound from client not in client's address range
# DONE ELSEWHERE
# Rule 3 for an ISP
# prevent entry of packets not destined clients from passing their way
# DONE ELSEWHERE
# Rule 4 for an ISP
# prevent exit of packets not from our class Bs
# on interface 0 (backbone) out filter
Interface 0 out
permit ip 123.7.0.0
permit ip 231.6.0.0
permit ip 201.96.0.0
deny ip 0.0.0.0
# Rule 5 for an ISP
# prevent entry of packets not destined for our class Bs.
# on interface 0 (backbone) in filter
Interface 0 in
permit ip 123.7.0.0
permit ip 231.6.0.0
permit ip 201.96.0.0
deny ip 0.0.0.0
Next, we implement rules 2 and 3 for each client by creating separate
(in this example ppp) filters on the ISP's gateway computer. Again,
using the Morningstar ppp Filter format and assuming that Class C IP
network 201.96.1.* is assigned to this particular client:
# Rule 1 for ISPs
# prevent known-bad address ranges from entering (or leaving)
!172.16-32.0.0 # private network segment
!192.168.0.0 # private network segment
!10.0.0.0 # private network segment
!127.0.0.0 # localhost network
# Rule 6 for an ISP
# prevent inbound traffic from the client destined for the client
# note that rule 6 is placed here because filters are order dependent
!recv/dest/201.96.1.0 # prevent inbound from client to self
# Rule 7 for an ISP
# prevent outbound traffic to the client claimed to be from the client
# note that rule 7 is placed here because filters are order dependent
!send/src/201.96.1.0 # prevent outbound to client from client
# Rule 2 for an ISP
# prevent inbound from client not in client's address range
recv/src/201.96.1.0 # allow legitimate traffic
!recv/src/0.0.0.0 # prevent all other traffic
# Rule 3 for an ISP
# prevent entry of packets not destined clients from passing their way
send/dest/201.96.1.0 # allow legitimate traffic
!send/dest/0.0.0.0 # prevent all other traffic
Note that redundant protection is provided in several ways. The ISP
protects the clients from backbone forgery both at the backbone router
and at the client's ppp connection, and protects the backbone from IP
forgery emanating from the ISP both by preventing forgery from clients
and by preventing forgery from within the ISP. Similarly, the ISP
provides redundant protection against improperly configured client
hardware and software. The last two filter rules are to assure that
even if the client is not properly configured to prevent forgery of
internal addresses from the outside world or to prevent internal
traffic from being sent out, this traffic is prevented.
This last example is a simplification of a wide area backbone network
in which this particular router (no type specified) is at the junction
between UK connections and non-UK connections. In this case, it is a
reasonable assumption that all internal UK traffic should remain
internal and that external traffic that gets to this node should be
sent out of the UK never to return. This particular backbone node will
be connected to non-UK traffic on interface 0, our previously
described ISP on interface 1, and the rest of the internal UK backbone
on interface 2.
# Rule 1 for a backbone
# prevent all of the known-bad packets from crossing
all-interfaces prevent in/out 172.16-32.0.0 # private network segment
all-interfaces prevent in/out 192.168.0.0 # private network segment
all-interfaces prevent in/out 10.0.0.0 # private network segment
all-interfaces prevent in/out 127.0.0.0 # localhost network
# Rule 2 for a backbone
# prevent packets originating from any ISP with non-ISP source address
interface-1 allow in from 123.7.0.0 # ISP traffic
interface-1 allow in from 231.6.0.0 # ISP traffic
interface-1 allow in from 201.96.0.0 # ISP traffic
interface-1 prevent in from 0.0.0.0 # no other inbound traffic
# Rule 3 for a backbone
# prevent packets not destined for an ISP from going there
interface-1 allow out to 123.7.0.0 # ISP traffic
interface-1 allow out to 231.6.0.0 # ISP traffic
interface-1 allow out to 201.96.0.0 # ISP traffic
interface-1 prevent out to 0.0.0.0 # no other outbound traffic
# Rule 4 for a backbone
# prevent packets from other backbones that shouldn't come this way
interface-0 allow in to UK-1 # UK traffic
interface-0 allow in to UK-2 # UK traffic
...
interface-0 allow in to UK-n # UK traffic
interface-0 prevent in to 0.0.0.0 # no other inbound traffic
# Rule 5 for a backbone
# prevent packets that should stay in our backbone from going out
interface-0 allow out from UK-1 # UK traffic
interface-0 allow out from UK-2 # UK traffic
...
interface-0 allow out from UK-n # UK traffic
interface-0 prevent out from 0.0.0.0 # no other outbound traffic
In this example, we have assumed that all UK traffic is on IP
addresses identified as UK-1, ..., UKn.
What-ifs and Objections
WHAT IFS?
* What if a private network ignores the rules? It is to be expected
than many private networks will ignore any such rules, either
through ignorance, intent, or inattention. But even if all private
networks ignored all of the rules, the rules for ISPs would
prevent IP forgery from extending to the overall infrastructure.
* What if an ISP ignores the rules? If an ISP ignores the rules and
allows IP forgery, the backbone can protect the rest of the
Internet, at least to the point where forged packets within the
ISP's domain remain within or are traceable to that domain. That
means that the ISP's clients would be subject to IP forgeries from
other clients of that ISP, but that the rest of the Internet would
be able to trace all packets coming from that ISP to that ISP.
* What if the backbone ignores the rules? If all of the backbone
providers ignore the rules, unless everyone else follows them, we
will continue to have IP forgeries through the ISPs that don't
follow the rules.
* What if combinations ignore the rules? Depending on the specific
combinations, we will have more or fewer IP address forgeries. It
turns out that a complete analysis of this issue is not simple
enough to do in the space provided, but a simple conclusion can be
drawn without a full analysis. As more Internet users and
providers prevent IP address forgery, the job of the forger will
become harder and harder. We don't all have to participate in
order to have proper protection, but if we all fail to
participate, the forgeries will continue.
OTHER OBJECTIONS
* Content (common carrier) objections: Many ISPs and backbone
providers don't want or take responsibility for content in the
Internet. Just like a telephone company, they don't want any role
in examining or dictating the content of the messages - they only
want to be a delivery service. It could be argued that examining
the address information in an IP packet and preventing packets
based on those addresses constitutes limitation of content. Of
course the portion of the content involved here must be examined
in order to route the information, and the routing used in the
Internet already provides exclusion of packets based on IP address
ranges. Furthermore, common carriers (in the U.S.) are allowed to
listen to and filter traffic to the extent that this activity is
done solely to assure the proper operation of the network. Thus
this objection would seem to be moot.
* The cost is too high objection: In fact the cost is negligible. If
the rules set forth herein were applied as a normal part of the
installation and maintenance process, it would come to only a few
minutes of effort during each installation. Even applying them to
systems already in place requires only a few minutes of effort,
again an insubstantial amount of effort well within the discretion
of any systems administrator.
* The we don't want restrictions objections: There are a substantial
number of people that want a total lack of restrictions on
information flowing through the Internet. I generally agree with
the principle of free information flow, except in cases where the
freedom of one person infringes on the freedom of others. This
impingement on other peoples' rights applies to certain types of
information, such as routing information, that must be correct in
order for the Internet to work properly. Since the restrictions
described here only assure that the Internet works properly and
don't restrict the content or flow of information, there is no
restriction of the free flow of information here. Only increased
assurance that those who want to use the media for legitimate
purposes will continue to be able to do so.
Summary
This solution we presented:
* 1) is easy to implement,
* 2) makes good sense from a traffic standpoint,
* 3) allows all legitimate activity without any hinderence,
* 4) works even if all parties don't participate,
* 5) costs almost nothing to implement at each site,
* 6) does not require any changes in existing protocols of traffic
patterns,
* 7) makes good sense for the security of each party that
participates, and
* 8) can be done today.
All that remains is for the people in each of these organizations to
implement these protections. Unlike so many of the problems in the
Internet that are hard to solve and will require years of evolution,
this problem can be solved now. We encourage you to implement these
protections at your earliest convenience and to urge other to do so as
well. Together, we can eliminate IP address forgery.