1. What’s “Inside” a Router?
Switching: the actual transfer of datagrams from
the router’s incoming links to it’s outgoing links.
4: Network Layer

2. Router Architecture Overview
Two key router functions:
run routing algorithms/protocol (RIP, OSPF, BGP)
switch datagrams from incoming to outgoing link
4: Network Layer

3. Input Port Functions Decentralized switching: Physical layer:
bit-level reception
Decentralized switching:
given datagram dest., lookup output port using routing table in input port memory
goal: complete input port processing at ‘line speed’
queuing: if datagrams arrive faster than forwarding rate into switch fabric (i.e., the packet is “blocked”)
Data link layer:
e.g., Ethernet
see chapter 5
Problem: how long does it take to perform a lookup?
4: Network Layer

4. Input Port Queuing
If routing fabric is slower than input ports combined -> queuing may occur at input queues
Head-of-the-Line (HOL) blocking: queued datagram at front of queue prevents others in queue from moving forward
queuing delay and loss due to input buffer overflow!
4: Network Layer

5. Three types of switching fabrics
(interconnection
network)
4: Network Layer

6. Switching Via Memory First generation routers:
packet copied by system’s (single) CPU
speed limited by memory bandwidth (2 system bus accesses per datagram)
Input
Port
Output
Memory
System Bus
DMA
Modern routers:
input port processor performs lookup, copy into shared memory
Cisco Catalyst 8500
4: Network Layer

7. Switching Via Bus datagram from input port memory
to output port memory via a shared bus
bus contention: switching speed limited by bus bandwidth (only one packet at a time can use bus)
1 Gbps bus - Cisco 1900: sufficient speed for access and enterprise routers (not regional or backbone)
4: Network Layer

8. Switching Via An Interconnection Network (commonly: Crossbar)
overcomes bus bandwidth
limitations
Banyan networks, other
interconnection nets initially
developed to connect processors
in multiprocessor design
advanced design: fragments
datagram into fixed length cells, switches cells
through the fabric.
Cisco 12000: switches up to 60 Gbps through the
interconnection network
4: Network Layer

9. Output Ports
Buffering required when datagrams arrive from the fabric faster than the transmission rate
Scheduling discipline chooses among queued datagrams for transmission
4: Network Layer

10. Output port queuing
buffering when arrival rate via switching fabric exceeds output line speed
queuing (delay) and loss due to output port buffer overflow!
Question: where is queuing
most likely to occur?
4: Network Layer

11. Cisco Catalyst 8500 Architecture*
Router Processor
8510: 100MHz R4600 RISC
8540: 200MHz R5000 RISC
Forwarding
Information Base
(routing tables)
Input Ports
4 ports/card
Output Ports
4 ports/card
Shared Memory
Switching Fabric
8510: 3 MBytes
8540: 12 MBytes
* From Cisco’s Web site
4: Network Layer

12. IPv6
Initial motivation: 32-bit address space completely allocated (gone!) by 2008 (maybe sooner… maybe later!).
Additional motivation:
header format changes were needed to improve speed of processing and forwarding
header changes were required to facilitate QoS
a new “anycast” address was needed: route to “best” of several replicated servers
IPv6 datagram format:
fixed-length 40 byte header (32 for addresses)
no fragmentation allowed
4: Network Layer

13. IPv6 Header (Cont) Priority: identify priority among datagrams in flow
Flow Label: identify datagrams in same “flow.”
(concept of “flow” not well defined).
Next header: identify upper layer protocol for data
ver.
class
flow label
8 bytes
32 bytes
4: Network Layer

14. Other Changes from IPv4 Fragmentation: not allowed/supported
Checksum: removed entirely to reduce processing time at each hop
Options: allowed, but outside of header, indicated by “Next Header” field
ICMPv6: new version of ICMP
additional message types, e.g. “Packet Too Big”
multicast group management functions (IGMP)
4: Network Layer

15. Transition From IPv4 To IPv6
Not all routers can be upgraded simultaneous
no “flag days” (e.g. NCP to TCP attempt in c. 1981)
How will the network operate with mixed IPv4 and IPv6 routers?
Two proposed approaches (RFC 1933):
Dual Stack: some routers with dual stack (v6, v4) can “translate” between formats
Tunneling: IPv6 carried as payload in IPv4 datagram among IPv4 routers
4: Network Layer

16. Dual Stack Approach
4: Network Layer

17. Tunneling
IPv6 inside IPv4 where needed
4: Network Layer

18. Multicast Routing (overview only)
unicast protocols: one sender and one receiver
multicast protocols:
one sender and a group of receivers
multicast abstraction: a single send operation to multiple receivers
Two approaches:
Sender sets
up multiple,
separate
unicast
connections
Explicit
multicast
support in
the network
layer
4: Network Layer

19. Multicast groups to D
1110
multicast address (28 bits)
address indirection: single Class D multicast IP address which identifies a group
multicast group: the group receivers, identified by specific host IP addresses, associated with a group
Class D multicast
address
4: Network Layer

20. Multicast routing IGMP message types: message format:
membership queries: general or specific
membership report: host wants to join
leave group: host leaves a specific group
message format:
Carried in IP datagram
IP protocol # of 2
Internet multicast routing algorithms:
MOSPF – open shortest path first
PIM – protocol independent Inter-AS
CBT – core-based trees
DVMRP – distance vector
Inter-autonomous routing:
DVMRP – de facto standard for multicast
PIM
4: Network Layer

21. Multicast Routing -Broadcast
4: Network Layer

22. Multicast Routing - Unicast
4: Network Layer

23. True Multicast (e.g. MOSPF)
4: Network Layer

24. Multicast Operation 4: Network Layer
Spanning Tree from Source Multicast Group
Packets Generated for Transmission
4: Network Layer

25. Multicast routing
Internet Group Management Protocol (IGMP): first and last hop protocol
wide area routing protocol: path through the network to multiple end systems, using network-layer multicast routing algorithms
4: Network Layer