IPv6

IP version 6 is the protocol which should replace contemporary Internet Protocol (denoted as IPv4). Its aim is to solve some problems caused by the Internet development (most notable the lack of IPv4 addresses) and provide prospective platform for further development.

CESNET has been working with IPv6 since 1999 when it launched the first experimental IPv6 backbone. Since 2004, we have provide the IPv6 as a common service, fully comparable to IPv4.

Topology of CESNET2 IPv6 network

We launched our first IPv6 testbed in 1999. Today, we are using the 6PE technology delivering IPv6 datagrams over MPLS therefore our IPv6 networks de facto does not have a separate topology. IPv4 and IPv6 datagrams are treated equally in our backbone.

Topologie IPv6 sítě

We are connected both to the IPv6 network of the GÉANT project and to production IPv6 Internet. Since 2004, we have been offering IPv6 in our backbone as a common service available at all points of presence.

Connectiong of network to CESNET2 over IPv6

Here we describe the options to choose and procedures to carry on when connecting an institution network to the CESNET2 backbone over IPv6. We do not deal with the local network issues in relation to IPv6.

Technical requirements the connected network has to fulfill

The network must be technically ready to transport IPv6. The readiness of the border router of the connected network is crucial for the connection to CESNET2 IPv6 backbone. We recommend to verify and update its firmware/software due to fast development on many platforms.

We verified in practice a connection using Cisco, Linux and *BSD PC routers.

Administrative steps

First, the institution has to ask CESNET NIC to get IPv6 address space assigned. The common policy is to assign /48 prefixes to connected end networks which allows 64 536 subnets of standard /64 length. Addresses are allocated from the prefix 2001:718::/32 with respect to the geographical location of the connected network.

Technical steps

Interconnection

CESNET2 backbone routers support both IPv4 and IPv6 (so called dual-stack). Interconnection is usually implemented over a single line for both protocols. Connections over a tunnel are not supported.

If the border router of the connected network is not dual-stacked, the organisation has to provide two border routers and connect IPv4 and IPv6 protocols separately. Interconnection to both routers can be implemented by a L2 switch. Technical solution should be consulted and adapted to capabilities of both sides.

Recommended and prevailing way is to integrate routing of both protocols into a single device (dual-stack). Major benefit of this approach is an easier architecture and management of connected network.

Routing

When configuring routing, in the first place it is necessary to ensure acceptability of address space on all border routers of connected network. Example for Cisco IOS:

ipv6 route 2001:718:1001::/48 Null0

This configuration guarantees processing of assigned address space inside the connected network and protect against routing loops for packets targeted to unused parts of assigned address space.

Routing of assigned IPv6 address space can be done statically or dynamically.

Static routing

If the connected network has got single point of connection (which is typical case), static routing is usually sufficient. In this case you just have to set up a default gateway on your border router to appropriate CESNET router.

Static configuration of default gateway on Cisco border router looks like this:

ipv6 route ::/0 2001:718:cafe:1::a1 50

Routing of institution address space on the CESNET2 side is also static.

Dynamic routing

If the connected network has got multiple connection points, dynamic routing using BGP is more appropriate. Autonomous system number is assigned by the CESNET2 management from private space. AS number is the same both for IPv4 and IPv6.

Essential BGP configuration in Cisco IOS looks like this:

router bgp 65050 bgp router-id 1.2.3.4 bgp log-neighbor-changes neighbor 2001:718:cafe:1::a1 remote-as 2852 neighbor 2001:718:cafe:1::a1 description CESNET IPv6 router address-family ipv6 neighbor 2001:718:cafe:1::a1 activate neighbor 2001:718:cafe:1::a1 soft-reconfiguration inbound network institution_prefix::/48 exit-address-family

Only the default gateway is propagated over BGP to the connected network. It is possible to propagate more destinations when agreed with CESNET2 network NOC.

Prefix lists or router maps must be added to BGP configuration on the institution side.

CESNET2 network routers accept in the BGP session only address space and AS number assigned to the institution. In the case of multipath connection, it is possible to influence incoming traffic by setting MED on corresponding BGP connections.

Current backbone load
IPv6 Addressing scheme