Skip to content

1.0 Network Principles

1.1 Identify Cisco Express Forwarding Concepts

  • IOS platform has 3 methods to forward packets (process switching, fast switching, CEF)
  • Layer 3 device has distributed architecture - independant control plane and data plane

    • Control plane (route processor)
      • runs routing protocol
      • conveys control info to interface module for control of data plane
      • collects data plane info (traffic stats)
      • handles certain data packets sent to the RP
    • Data plane (interface microcoded processor)
      • forwards data packets
  • Process switching (slowest)

    • Each packet is individually examined by the CPU (control plane)
    • All forwarding decisions made in software
    • CPU intensive/last resort due to degradation of performance
  • Fast switching (faster)
    • Initial packet in a flow is process switched
    • Forwarding decision is stored in the data plane's hardware fast-switching cache
    • Subsequent frames are forwarded based on the fast-switching cache
    • Fast-switching cache contains Address, Prefix, L2 mac of next hop, outgoing interface
  • Cisco Express Forwarding (fastest)
    • Least CPU intensive
    • Control plane creates two hardware-based tables
      • Created using Layer 2/3 tables (incl. routing and ARP tables)
      • Contain all info needed to forward packet
      • Hardware based decision for all frames in flow
      • CEF caches info from routing engine before flows are encountered
      • Control plane is responsible for building FIB/adj tables
      • 1.1.a Forwarding Information Base (FIB) show ip cef show ipv6 cef
        • Contains precomposed reverse lookups and next-hop information
        • Contains Address, Prefix, 'Adjacency pointer' (to Adj table)
        • Dest prefixes are stored from most specific to least specific (maxiumum efficiency)
      • 1.1.b Adjacency table show adjacency
        • Holds Layer 2 next-hop addresses and frame header rewrite info for all FIB entries
        • Contains 'Adjacency pointer', L2 mac of next hop (from ARP table)
        • Populated as adjacency's are discovered (via ARP)
    • Some features are not compatible with CEF
      • IP header options
      • expiring TTL counters
      • forwarded to tunnel interface
      • unsupported encapsulation (in or out)
      • Exceed MTU and require fragmentation
    • Pakcets that cannot be CEF switched are 'punted' to be fast/process switched
    • Enabled by default for IPv4, enabled with ipv6 unicast-routing for IPv6

! validate that CEF is running
show interface Fa0/0
 ..
 IP CEF switching is enabled
 ..
(config-if)# no ip route-cache cef disable CEF per interface (config)# no ip cef disable CEF globally

1.2 Explain general network challenges

a. Unicast - Traffic exchanged between one sender and one reciever b. Out-of-order packets c. Asymmetric routing - Traffic flows that use a different path for the return path than the original - Occurs in networks w/ redundant paths - Often desirable for effective bandwidth use


1.3 Describe IP operations

a. ICMP Unreachable and Redirects - IMCP Redirect: used by routers to notify sender that there is a better route available - If a router recieves a pkt from segment, and sees the destination on the same segment, it will notify the sender with an ICMP Redirect to packets can be forwarded directly

b. IPv4 and IPv6 fragmentation - IPv4 packet larger than egress MTU - must fragment unless DF bit is set - Causes CPU/memory overhead during fragmentation/reassembly - If one fragment is dropped -> retransmit whole packet - L4/L7 filtering has trouble processing - - IPv6 fragmentation - IPv6 routers only fragment a packet if they are the source - Packets larger than the MTU of outgoing interface are dropped - Sends ICMPv6 'Packet Too Big' message back to the source, including the smaller MTU

c. TTL


1.4 Explain TCP operations

a. IPv4 and IPv6 (P)MTU - IPv4 packet max size: 65,535 bytes - IPv6 packet max size: 4,294,967,295 bytes (w/ hop-by-hop extension and jumbo payload) - Most transmission links enforce smaller sizes

b. MSS - defines largest amount of data a recieving device can accept in a TCP segment - manually set -> not negotiated - MSS should = MTU - 40 bytes (20 for IPv4 header, 20 for TCP header) - Helps avoid fragmentation but does not prevent it if there is a smaller MTU link in the path

  • PMTUD (Path MTU Discovery)
    • determines the lowest MTU along a path
    • supported by TCP
    • Requires ICMP Destination Unreachable messages
      • Ensure that 'unreachable' and 'time exceeded' messages are permitted by packet filtering
    • PMTUD works for IPv6 similar to IPv4

c. Latency - amount of time for a message to go from A to B - Network Latency: amount of time for the packet to travel the network from src to dst - Causes of latency - Propagation Delay - Serialization - Data Protocols - Routing - Switching - Queueing - Buffering - TCP flow control & reliability have an affect on end-to-end latency

d. Windowing - "the amount of data that can be sent before expecting an acknowledgement" - Usually several times the MSS

e. Bandwidth-delay product (BDP) - TCP can experience bottlenecks on paths with high bandwidth and long RTT (i.e. long fat pipe) - BDP = bandwidth (bps) x RTT (seconds) - BDP = "number of bits it takes to fill the pipe" i.e. "amount of unacknowledged TCP data to keep the pipeline full" - BDP is used to optimize the TCP window size

f. Global synchronization


1.5 Describe UDP operations

a. Starvation - aka "TCP starvation" or "UDP dominance" - TCP - includes reliability, flow control, congestion avoidance - UDP - includes none of the above - When TCP+UDP flows experience congestion, TCP backs off using 'slow start' - UDP then eats up the bandwidth that TCP gave up

b. Latency - UDP has very low latency compared to TCP - UDP - used for streaming media that requires minimal delay and can tolerate occasional packet loss


1.6 Recognize proposed changes to the network

a. Changes to routing protocol parameters b. Migrate parts of the network to IPv6 c. Routing protocol migration


References

https://www.cisco.com/c/dam/en_us/training-events/exams/docs/300-101_route.pdf