I Presentation by ANML

June 2004
N
D
I
A
N
A

U
N
I
V
E
R
S
I
T
Y
IPv6 and Security
I
N
D A Bold Statement
I
A
N
“…our soldiers need better information in order
A
to make better decisions—who to help and
who to kill. The lack of security and flexibility
U in the current IPv4 protocol is a drag on our
N wing. This isn't about do you trust the Internet
I for your kid's homework, it's do you trust your
V kid's life. If we fail, people die.”
E – Defense Department Will Require IPv6
R Compliance, Says DoD's John Osterholz. From
S Market Wire, June 26, 2003.
I
T
Y
I
N
D What is This Added Security?
I
A
N
A

U
N
I …(not much)…
V
E
R
S
I
T
Y
I
N
D Overview
I
A
N • That was another bold statement (and we
A
had better back it up!)
U • We will focus first on the differences in
N
security between IPv4 and IPv6, then on the
I
V
similarities
E • This will include some social arguments
R
S
I
T
Y
I
N
D IPsec Support
I
A
N • The major difference is that an IPv6 device
A must support IPsec
– IPsec is available for both IPv4 and IPv6 but is
U
N
not a requirement for IPv4
I – Configuration and use is functionally identical
V between IPv4 and IPv6
E
• This still leaves the question of when to
R
S
actually use IPsec
I
T
Y
I
N
D IPsec Support
I
A
N • IPsec is difficult to install and configure on most
A platforms
– This is especially true with the retirement of the
U
FreeS/WAN project
N
I
• Biggest problem is key distribution
V – Requires infrastructure support (e.g., special DNS RRs)
E and dedicated professional staff
R – If you get this part wrong, you gain complexity without
S any additional security
I
T
Y
I
N
D IPsec Adoption
I
A
N • IPsec has been around since 1995, but still
A
sees limited use outside of L2TP-based
U VPNs
N
• Why?
I
V – Much more ubiquitous support for SSL
E – IPsec and NAT don’t mix well at all
R
S
I
T
Y
I
N
D SSL vs. IPsec
I
A
N • IPsec is better than SSL because it provides
A much better protection for packet headers
– Provides confidientiality, accountability, and
U
N
authentication
I – No more spoofed headers, etc.
V • SSL is better than IPsec because you have it
E
right now and it works pretty well for just
R
S
about everything you want to do
I
T
Y
I
N
D Won’t NATs Go Away?
I
A
N • Part of the purpose of IPv6 is to restore the end-
A
to-end model by providing more addresses
– But address depletion is not the only motivating force
behind NATs
U
– Security practices are at least as much to blame
N
I
• NATs probably provide the best cost-to-benefit
V
ratio of any simple security measure
E
– A NAT box is dirt-cheap and easy to configure
R
– It also completely breaks the end-to-end model
S • There will still be NATs in IPv6
I
T
Y
I
N
D Address Sparsity
I
A
N • Many IPv4 worms and cracking tools do scans of
A IPv4 address space to find hosts
• IPv6 increases the size of the address space by
U
N
over 79,000,000,000,000,000,000,000,000,000
I times
V – Properties of the address structure can pare down the
E search space somewhat
R – Nevertheless, brute force search of IPv6 address space
S will be completely intractable
I
T
Y
I
N
D Does This Gain Us Security?
I
A
N • It does eliminate a primary technique of a
A great deal of malware (and some legitimate
research efforts)
U
N • Lists of hosts to attack will be harvested
I from system configuration files, e-mail
V addresses, Web sites, server logs, etc.
E
R
– This is exactly how the Morris worm worked
S
back in the late 1980s
I
T
Y
I
N
D Does This Gain Us Security?
I
A
N • How well-known will a host need to be before its
A address leaks into this lists?
– How much spam do you get?
U
N
• There is a bright side to this:
I – A long list of addresses takes up a lot of space and
V provides forensic evidence
E – You won’t have packet-of-death attacks like SQL
R Slammer any more
S
– Worms are more likely to report back to their source
I
T
Y
I
N
D Exposure of MAC Addresses
I
A
N • Standard IPv6 addresses contain the MAC
A
address in the lower 64 bits of the addresses
U – This is information that was usually confined to
N a single broadcast domain before
I
– The manufacturer of your NIC is now public
V
E
knowledge and may associate you with a
R known vulnerability
S
I
T
Y
I
N
D Things That Stay the Same
I
A
N • IPv6 doesn’t change TCP or UDP at all
A
• IPv6 doesn’t patch vulnerabilities in
U individual applications or OSes
N
I
• IPv6 doesn’t force network administrators
V to do egress filtering
E
R
• IPv6 doesn’t mandate use of any security
S features
I
T
Y
I
N
D Things That Stay the Same
I
A
N • A recent survey of CERT’s top 100
A
vulnerabilities shows only one of them to be
U specific to IPv4
N – You’ll still have e-mail trojans, buffer overruns,
I
macro viruses, and so forth
V
E – You’ll also still have SYN floods, RST attacks,
R OS fingerprinting, hijacked connections, and
S the like
I
T
Y
I
N
D Code Maturity
I
A
N • Most of the IPv6 code in the world is new and
A untested in comparison to IPv4
• This code is certain to contain more flaws and
U vulnerabilities than its IPv4 equivalents
N
– It’s larger and much more complex
I
– It has not yet stood the test of time—or attacks
V
E
• This situation will slowly improve over time
R – IPv6 isn’t low-hanging fruit yet, so there’s little
S motivation to attack it
I
T
Y
I
N
D Code Maturity
I
A
N • Flaws will be opened in existing applications as
A they are ported from IPv4 to IPv6
– IPv6 involves many more programming changes than
U
just bigger addresses
N
– Net increase in lines of line
I
V • New code will be written to deal with (or reinvent)
E third-party libraries that do not handle IPv6 and
R cannot be modified
S
I
T
Y
I
N
D Protocol Maturity
I
A
N • Not only is IPv6 code comparatively immature,
A but so are its standards
• Example: A6 vs. AAAA DNS records
U
N – A6 was clever, but raised concerns about DoS attacks
I using infinitely recursive delegated lookups
V – Now relegated to experimental status
E
• Similar concerns have been expressed about
R
protocols for tunneling IPv6 in IPv4 networks
S
I
T
Y
I
N
D Protocol Maturity
I
A
N • Many features are not fully specified yet
A
– What do you do with the Priority field?
U – What’s the exact structure of the Flow Label?
N
– What’s the format of Aggregatable Global
I
Unicast Addresses this year?
V
E – When and how often do I do MTU discovery?
R – How do anycast addresses actually work?
S
I
T
Y
I
N
D Router Maturity
I
A
N • Issues with code and protocol maturity come
A
together in the routers
– A vulnerable host may result in the loss of a single
system
U
– A vulnerable router may result in the loss of a
N
substantial piece of the network
I
• Catch-22: Router vendors can’t spend too much
V
on testing IPv6 stacks until IPv6 gets more
E
popular, and IPv6 has a hard time getting more
R popular until router vendors spend more time
S testing their IPv6 stacks
I
T
Y
I
N
D Competing and Complex Standards
I
A
N • In some ways, IPv6 suffers from “design by
A committee” spread across multiple committees
• The IPv6 Address Oracle has to draw from over a
U
N
dozen different RFCs
I • Examples of multiple standards
V
– DNS: AAAA vs. A6
E
– Tunnels: At least four different approaches
R
S
– Resolver: getipnodebyaddr() vs. getaddrinfo()
I
T
Y
I
N
D Best Practices
I
A
N • Be prepared to devote considerable resources to
A development and maintenance of key
infrastructure if you plan to use IPsec
U • Adopt new features of IPv6 sparingly until their
N
standards processes are finalized
I
V
• Allow for the existence of more undiscovered
E
flaws in IPv6 code when assessing risks
R • Subject ported applications to the same level of
S review and testing as new ones
I
T
Y
I
N
D Best Practices
I
A
N • Have clear definitions of “IPv6 ready” and “IPv6
A aware” when you compare vendors’ products
• Pay close attention to new RFCs as they come
U
N
out—and changes in the status of old ones
I • Design new protocols in such a way that they will
V continue to operate through a NAT
E
R
• Don’t write IPv6-only applications; make them
S dual-stack instead
I
T
Y
I
N
D Conclusion
I
A
N • IPv6 does not make for a completely
A
different world of security
U • Expect a low level of incidents initially
N
(obscurity), followed by a much higher
I
V
level (exploitation), followed by a slow
E decline to the level we see now with IPv4
R
(stasis)
S
I
T
Y