Download Core of the network

Tomáš Podermański, [email protected]
• the Czech Republic
–
–
–
–
–
–
Area: 78 866 Km2
Population: 10 230 060
Capital city : Prague
1989: communist regime collapsed
1999: meber of NATO
2004: member of EU
Brno
NREN – CESNET z.s.p.o.
– 26 members
– Universities & Czech academy
– Many other institutions connected indirectly
– Government institutions, hospitals, high schools,
research institutions
– Network based on DWDM & MPLS technology
Rumors about IPv6
• Do you know that IPv4 address space will be
exhausted very soon ?
– Really, again? I already heard about it 10 years ago.
• We are working on IPv6 transition !
– Do you still use it?
• How do you get on with IPv6 ?
– Everything is ready. The network is well prepared and all
applications support dual stack. There is no problem with IPv6.
• Very well, can I see how it works?
– Ehh, em, you know…. maybe later
• Actual point of view
– We know that IPv6 is a real problem, but we don’t have time and
money to deal with it.
– We’d like to cope with IPv6, but we don’t know how.
Why IPv6 ?
Top 10 Features that make IPv6 'greater' than IPv4
1. Larger IP address space
2. Better end-to-end connectivity
3. Ability for autoconfiguring devices
4. Simplified header structures
5. Better security (IPSEC – ESP, AH)
6. Better quality of services
7. Better multicast and anycast abilities
8. Mobility features
9. Ease of administration
10.Smooth transition from IPv4
source: http://ipv6.com/
Why IPv6 ?
Top 10 Features that make IPv6 'greater' than IPv4
1. Larger IP address space
2. Better end-to-end connectivity
3. Ability for autoconfiguring devices
4. Simplified header structures
5. Better security (IPSEC – ESP, AH)
6. Better quality of services
7. Better multicast and anycast abilities
8. Mobility features
9. Ease of administration
10.Smooth transition from IPv4
source: http://ipv6.com/
Ten years ago we had a plan …
IPv4 Pool Size
IPv6 Deployment
Size of the
Internet
IPv6 Transition using
Dual Stack
6 - 10 years
Time
Source: http://www.potaroo.net/presentations/
What’s the revised plan?
IPv4 Pool
Size
100
%
Size of the
Internet
?
Today
1%
IPv6 Transition
IPv6 Deployment
1 year
Time
Source: http://www.potaroo.net/presentations/
An Internet Transition Plan
• Recommendation based on RFC5211 (07/2008)
• Phase I (2008 – 2009)
– Backbone network, basic infrastructure
– Native connectivity to each location (low speed)
– Some public services available on IPv6 (web, ftp)
• Phase II (2010/1 - 2011/12)
–
–
–
–
–
Stable backbone infrastructure
Hardware routing
Monitoring of applications & hotline support
IPv6 connectivity for end users in selected locations
IPv6 multicast in testing mode
• Phase III (2012/1)
– IPv6 network in good working order (unicast & multicast)
– Native IPv6 connectivity for all users
– majority of services available through IPv6
IANA: Allocation of /8 IPv4 prefixes
2008: Google on IPv6
2008: European commision
targets 25% IPv6
availability by 2010
NAT
1996: IPv6 pon Linux
2007: Win Vista,
SP1 dulastack
Prefered IPv6
1998: RFC 2460
1992: The first
call
Source: http://www.potaroo.net/tools/ipv4/index.html
2001: Win XP,
SP1 dulastack
Action plans of deployment IPv6
• EU : ADVANCING THE INTERNET
– Action Plan for the deployment of Internet Protocol version 6 (IPv6) in
Europe,Brussels, 27.5.2008, COM(2008) 313 final
• USA: Transition Planning for Internet Protocol Version 6
(IPv6), to set the US Federal Agencies a hard deadline for
compliance to IPv6 on their core IP networks
• China : China Next Generation Internet (CNGI) sets out a 5 year plan
(2006-2010) for the early adoption of IPv6
• Korea: IPv6 Promotion Plan II which sets a vision of deploying IPv6
for the public sector by 2010
• Australia: Preparation Jan 2008-Dec 2009, Transition Jan 2010-Dec
2012, Implementation Jan 2013-Dec 2015
Google: users with IPv6 connectivity
Source: http://www.google.com/intl/en/ipv6/statistics/
IPv6 deployment ratio – 12/2010
Source: http://bgpmon.net/blog/?p=340
NREN‘s activities
• Backbone infrastructure
– IPv6 has been available in each node for
many years
– Every member can have an IPv6
connectivity
• IPv6 workgroup
–
–
–
–
meetings 3 times a year
a few campuses are usually presented
Usually 20 – 30 participants
good place to
• meet specialist from other universities
• share experience & knowledge
IPv6 and Czech Universities
Let’s move on to the campus
IPv6 status at
the Brno University of Technology
the Brno University of Technology
•
•
•
•
•
•
•
http://www.vutbr.cz
One of the largest universities in the Czech Republic
Founded in 1899, 110th anniversary was recently celebrated
20,000 students and 2,000 employees
9 faculties
6 other organizational units
Dormitory for 6,000 students
VUT FP, FEKT, Kolejní 4
VUT Koleje, Kolejní 2
VUT FCH, FEKT, Purkyňova 118
VUT Koleje, Mánesova 12
VUT FEKT, Technická 8
VUT FIT, Božetechova 2
VUT FSI, Technická 2
AV VFU, Palackého 1/3
VUT TI, Technická 4
VUT Koleje, Purk.
MU CESNET , Botanická 68a
AV ČR UPT
MZLU, Tauferova
VUT, Kounicova 67a
VUT Koleje , Kounicova 46/48
AV ČR UFM
VUT Rektorát, Antonínská 1
VUT FAST, Veveří 95
VUT FaVU, Údolní 19
VUT , Gorkého 13
VUT FEKT
Údolní 53
VUT FA, Poříčí 5
MU, Vinařská 5
VUT FaVU, Rybářská 13
AV ČR, Rybářská 13
Milestones
1992-1995
Modems, dedicated connections, bandwidth 32 - 128 Kb/s, first fiber was build, Ethernet 10 Mb/s, PC
based routers KA9Q a BSD/386.
1995-1998
connections among locations almost transferred to fiber, the ATM 155 Mb/s was being built, PC based
routers with BSD Unix.
1998-2001
optical connections with multiple fibers, first attempts to build up circuits on Gigabit Ethernet, L3
switches Extreme Networks.
2002-2004
All-core-network circuits on backbone converted to Gigabit Ethernet. Gigabit Ethernet was used to
connect each location.
2005
First experience with 10 Gbp/s Ethernet. Looking for popper technology to build up new backbone.
2006-2008
Step-by-step converting from Gigabit Ehernet to 10 Gigabit Ethernet. Selected technology: Extreme
Networks, Hewlett Packard.
2008
For management and L2 cross connections the backuped L2 circuit has been
build. Each location had been connected on 10 Gigabit.
2009-now
The IPv6 backbone has been built up. All tunnel-like connections have been
converted to native. The OSPFv3 based routing has been turned on. IPv6
connectivity is now ready to be used in all locations.
Layer 3 network
Core of the network
• Based on 10Gb/s ethernet
• Basic L3 services
• OSPF and OSPFv3
• multicast - PIM/SM
External connectivity
• Two 10Gb/s lines connecting the
core to CESNET (BGP, BGP4+)
• Basic filtering (SMTP, NetBios,
445/Microsof DS)
Locality & sub-campuses
• Two 10Gb/s lines to the core
• More complex firewalls
configurations are dependend on
local administrators
Initial IPv6 topology
IPv6 milestones
2002
Basic tunneled connectivity. Assigned own prefix - 2001:718:802::/48 .
2002-2008
Some experimental services. Possibility to connect locations using IPv6 (VLANs) . Static routing
based on FreeBSD PC routers.
Native connectivity to NREN
2009
Address plan, prefix divided into organization units.
OSPFv3 based routing. PC routers with XORP.
3com 4800 GL devices used as HW routers
DNS server moved to the dualstack
2010/I, 2010/II
Backuped connectivity to each location
Every place/subnet can support native IPv6 connectivity
Tests with HP devices (participation on beta testing program)
Connectivity to NREN through two 10Gb/s lines – BGP4+
Basic firewall
Monitoring of IPv6 services, collecting neighbor caches (NAV)
Some services moved to dualstack
Participation on HP beta testing program
2010/III, 2010/IV
Core of the network moved to the dualstack
Ridding off the temporary IPv6 network
IPv6 milestones - future
• Firmware with full IPv6 support has been released
– Temporary solution on xorp routers can be switched off
– IPv6 topology will follow the IPv4 topology
– All subnets will have both IPv4 and native IPv6 connectivity
• PI IPv6 address range has been assigned
– Changing address of all subnets and services. We will move from
2001:718:802::/48 to 2001:67c:1220::/46
– Support for renumbering
• Activation of services on dualstack
– 90% of services could be moved easily (10% of time)
– rest of services (90% of time)
• very complicated issue
• unpredictable problems
IPv6 problem no. 1
Autoconfiguration
Autoconfiguration
• Brand new autoconfiguration mechanisms
– Router advertisement (doesn’t contain address of DNS servers)
• There is an extension RFC 6016 but is not widely implemented yet
– DHCPv6 (doesn’t contain default route option)
• There is an draft draft-ietf-mif-dhcpv6-route-option-02 but not accepted yet
• Privacy extensions
– IPv6 addresses are created by hosts randomly
• You have to use both mechanisms in real network
– DHCPv6 server, Advertises on router
– Secure either of them
– Failure any of them leads to a network connectivity failure
• Different platforms support different techniques
– Windows XP - SLAAC
– Windows Vista/7 – SLAAC + DHCPv6
– MAC OS, iOS - SLAAC only (expect Lion – released 06/2011)
– Linux, BSD, … – depends on distribution
Autoconfiguration IPv4 x IPv6
• IPv4 – DHCP, ARP
• IPv6 – DAD, RS/RA, DHCPv6, MLDv2, ND
IPv6 problem no. 2
Impact on existing IPv4 infrastructure
IPv4 and IPv6 in a network
• More than 50% of PC supports dualstack
– Most of them use autoconfiguration (SLAAC) to get IP
address (MS Vista/7, Linux, Mac OS, iOS, BSD*)
– IPv6 is preferred protocol by default
• Steps to make an attack:
– Setup attacker’s IP to act as a RA sender
– Prepare a DHCPv6 server on the attacker's PC; as DNS
servers provide attacker’s addresses
– Modify the behavior of DNS server to return A or AAAA
records for www.google.com, www.yahoo.com, etc. to
your attacker’s address
– Transparent proxy service allows attacker to modify
content of webpages
IPv6 man in the middle attack
Port security:
• MAC address security
• DHCP snooping
• ARP protection
• Dynamic lock down
www.vutbr.cz
147.229.2.15
IPv6 man in the middle attack
Rouge Router Advertisement with M or O flag enabled
Rouge IPv6 Router
www.vutbr.cz
147.229.2.15
IPv6 man in the middle attack
Rouge DHCPv6 Server
www.vutbr.cz
147.229.2.15
DHCPv6 query (via multicast)
IPv6 man in the middle attack
DHCPv6 answer
DNS servers points to ME
Rouge DHCPv6 Server
www.vutbr.cz
147.229.2.15
IPv6 man in the middle attack
192.168.1.166
- name server
- proxy service
www.vutbr.cz
192.168.1.166
# ./flood_router6 eth0
IPv6 problem no. 3
First hop security
Autoconfiguration – IPv4
• IPv4 autoconfiguration = DHCP
• Protection mechanisms on L2 devices
– DHCP snooping
• Blocking DHCP responses on access ports
• Prevents against fake DHCP servers
– Dynamic ARP protection
• MAC-IP address database based on DHCP leases
• Checking content of ARP packets on client access port
• Prevents against ARP spoofing
– Dynamic lock down
• The MAC-IP database is used for inspection of client source MAC and
IP address.
• Prevents against source address spoofing
Possible solutions for IPv6
• SeND (RFC 3971, March 2005)
– Based on cryptography CGA keys
– Requires PKI infrastructure
– Can not work with
• Manually configured, EUI 64 and Privacy Extension addresses
• RA-Guard (RFC 6105, February 2011)
– Dropping fake RA messages on access port (RA Snooping)
– Cooperation with SeND (send proxy) – learning mode
• SAVI (draft-ietf-savi-*, divided into more drafts)
– Complex solution solving
• Rouge RA, DHCPv4 an DHCPv6
Possible solutions for IPv6
• No support in devices
– Only few vendors support some of that features
– You probably will have to replace all access switches
• There is a easy way how to bypass such
protection
– CVE-2011-2395
(http://seclists.org/fulldisclosure/2011/May/446)
– Using extension headers
– ICMPv6/ND Packet fragmentation
IPv6 problem no. 4
Impact on existing IPv4 infrastructure II
Impact on existing IPv4 infrastructure
• Deploying IPv6 requires changes in the existing
infrastructure
– Some routers and switches have to be replaced
– Some have to be upgraded
– There are a lot of bugs in IPv6 code
• More often firmware upgrades
• New IPv6 related bugs -> CPU overload, Devices crash
• IPv6 is another way how to reach nodes inside of
the network
– Similar policy have to be applied on both protocols
• Users (and managers) can see that network is
less stable
• Both that group can’t see any benefits of that effort
IPv6 problem no. 5
IPv6 can’t provide same services as IPv4 does
IPv6 can’t provide same services
• Implementations of IPv6 are very immature now
– Lack of VRRP (v3 standardized 1 year ago)
– DHCPv6 failover is neither standardized nor
implemented yet
– Issues on VPN
• Most of vendors supports only IPv4 over VPN
• What happened when user connected via VPN tries to reach
some services available on both protocols ?
– Implementations do not provide equal features
• ACL, OSPFv3, BGP4+, policy routing, …
– There are a lot of bugs in implementation
• Deploying IPv6 means that you will have to help supplier to
solve issues.
• Nobody else may have not seen some issues before
IPv6 problem no. 6
IPv6 requires brand new techniques to manage
networks
New techniques to manage networks
• Privacy extension
– Addresses are created randomly, a device has multiple address
– Collecting ND cache (NAV - http://metanav.uninett.no/)
• Netflow data have to be threaten in differed way
– Practical IPv6 Monitoring on Campus
(http://www.terena.org/activities/campus-bp/bpd.html)
• DHCPv6 uses DUID as host identifier
– MAC address can’t be used as host identifier
– The host can change DUID - OS reinstallation, dualboot systems
• Absence of NAT
– There are some scenarios where NAT is very useful
– RFC 6296 - IPv6-to-IPv6 Network Prefix Translation
• Tunnels
– An easy way to bypass security policy
IPv6 problem no. 7
Deploying IPv6 does not mean that you can
switch off IPv4
Switching off IPv4 is not easy
• IPv4 and IPv6 are incompatible protocols
– Services available on IPv4 can be reached only by IPv4
– Applications not supporting IPv6 still needs IPv4
• Skype, ICQ, MSN, games, …
• NAT64 in not solution for that problem today
– It is more realistic keep running dual stack network
• It means that you have to run
– Two routing protocols including routing policy
– Two autoconfiguration mechanism
– Two security policy
• IPv4 does not have extension headers but IPv6 does
• IPv6 uses differed messages in ICMPv6
• IPv6 can use multiple addresses
– You have to have well trained staff with very good knowledge of
both protocols
IPv6 problem no. 8
Administrators gets disappointed and frustrated
with developing IPv6
Deploying IPv6 in reality
• How things an administrator that has never heard
about IPv6
– Is it easier than IPv4 ?
• SLAC – security issues, identity problems
– Is it the same as IPv4 ?
• Static address configuration, link-local address
– Is it similar to IPv4 ?
• DHCP(v4) + DHCPv6 + RA, …
– There are too many differences
• There is to many new protocols, new mechanisms, …
• I have to learn too many new thinks
• Users will not appreciate it
A bit of statistics…
collected from the campus network
IPv4, IPv6 & tunneled traffic
Here
IPv6 native traffic
How many addressed were we talking with
Here
How IPv6 many addressed were we talking with
Come
H on, what all this
fuss is about. Just take it
easy and see what will
happen.
Number of unique mac adresses in ARP and NC
What can we do about it ?
• Start using IPv6 immediately
– We have been waiting for perfect IPv6 more than 15 years - it does
not work
– Until IPv6 is used we will not discover any problem
• Prefer native IPv6 connectivity (anywhere you can)
– It is a final solution for future (IPv4 will be switched off later)
– Native IPv6 is more secure than unattended tunneled traffic !
• Ask vendors and creators of standards to fix problems
– More requests escalate troubles on the vendor side
– Standardization of IPv6 is not enclosed process. Anyone can
contribute or comment the standards
• Stop pretending that IPv6 does not have any troubles
– IPv6 has got many problems
– Problems can not be solved by covering them
– Unreliable information led to broken trust amongst users. The
naked truth is always better than the best dressed lie