中文网站
  Advanced Search
Read the latest Blogs from IT professionals in the field. Read and write community created documents. Need IT help? Ask our staff. Connect with your peers. Check our Tech Shop for posters, books and software tools. Home

12.2.1 The SURFnet5 Dual Stack network

The SURFnet5 network consists of a core that is situated at two locations in Amsterdam, at the Points of Presence (PoPs) called “Amsterdam1”1 and “Amsterdam2”2. Two Cisco 12416 routers are installed at each location implementing the core functionality. Fifteen concentrator PoPs are each connected to both core locations over two separate unprotected SONET/SDH framed lambdas running at 10 Gbit/s. The engineering of the lambdas is such that it ensures that one connection is always maintained in case of a single transmission or core router failure. At each of the fifteen PoPs a router of type Cisco 124xx is installed together with a Cisco 7507 router for the low speed, legacy connections at speeds up to 155 Mbit/s. The backbone of SURFnet5 makes use of IP-over-DWDM, using POS framing. BT owns and operates the DWDM transmission infrastructure. Figure 12-4 shows the logical topology of SURFnet5.

From the start of SURFnet5 all routers running Cisco’s IOS have IPv6 implemented. During the early days of the network, IOS versions in the 12.0ST train were used in the GSR routers. During 2002 the transition to 12.0S was made. The current version, in March 2004, in the GSR routers is “IOS Version 12.0(26)S1” and in the 7500 routers is “IOS Version 12.2(14)S2”.

SURFnet5 started as a dual-stack network and has been running as such until March 2004. During March 2004 IPv6 routing within the core of the network was migrated from dual-stack to 6PE. The reason for this migration was to achieve line rate IPv6 forwarding in the SURFnet5 network. SURFnet5 is a 10 Gbit/s network largely build on Engine4 line cards and these cards handle IPv4 on the fast path while IPv6 is handled by the processor of the line card. After a large replacement in 2003 of Engine2 line cards with Engine3 line cards, the edges of SURFnet5 became capable of handling IPv6 traffic at line rate. By implementing 6PE in the core of the network it was ensured that IPv6 traffic was also handled at line rate in the core and potential bottlenecks were removed.

Figure 12-4 Logical topology for SURFnet5

The routing setup as implemented and running today runs without problems. A number of features are on SURFnet’s requirements list, of which IPv6 multicast is high on the list. SURFnet already has hands-on experience with IPv6 multicast in the 6NET context using external MBGP for IPv6.