Tdc 560 - Lecture 1

Front 1

A label switch patch is also referred to as a

Back 1

extremely flexible tunneling mechanism that allows a network IP routers to transport a packet to any other location using LSP (layer Switched Path) which is essentially a Layer 2 tunnel that is not routing of the packet involed

Front 2

How is LSP different from routing?

Back 2

Instead of looking up the IP destination in the routing table it will look up the label in a LFIB

Front 3

What is the general technique that is a pwer mechanism for implementing various service in the MPLS network?

Back 3

label switching

Front 4

Why does an MPLS network offer stronger QoS capabilities?

Back 4

Because of traffic engineering, pre-planning of traffic flow, and load balancing that provide tighter guarnatees instead of leaving to IPv4 routing

Front 5

What are internet hops referred to as

Back 5

Subnets

Front 6

What are the types of subnet

Back 6

Subnet can be:
a) Point-to-point physical link (2 IP addrs, one each end)
b) Layer 2 virtual circuit (i.e. Frame Relay, ATM, MPLS)
(2 IP addrs)
c) A LAN (many subnet IPs and possibly > 2 routers
connected to subnet)

Front 7

What is an IP subnet?

Back 7

What is an IP Subnet?
•All IP devices on same subnet can communicate with each other without going through a router.
–IP addresses of devices within same subnet start with same common Subnet Prefix.
–Any IP device can send broadcast to all devices in same subnet (but not usually to other subnets).

Front 8

What is used to interconnect different subnets?

Back 8

Routers interconnect different IP subnets.
–Each router interface connects to a different IP subnet
–Routing stores IP address and Subnet Mask for each port, which together specify all IP addresses in the subnet.

Front 9

What is a Subnet ID and what is the form?

Back 9

Subnet IDs
•All IP addresses in subnet can be represented by a Subnet ID of the form
–<Routing Prefix> / <prefix length>
•Where
–<Routing Prefix> is an IP address where all Host bits (last (32-<prefix length>) bits) are equal to 0
–<prefix length> is number of common prefix bits for all IP addresses in the subnet.
–The broadcast address for the subnet is the Routing Prefix with all Host bits set to 1
–<Routing Prefix>+1 is first useable IP address
–<Broadcast> - 1 is the last useable IP address

Front 10

Interface Configuration
•Combination of IP address and subnet mask informs the router about every IP address that is directly reachable through that interface, not just one.
•For example: What does this config tell the router?
interface s0/2/0
ip address 10.200.1.5 255.255.255.0

Back 10

It tells the router 2 things:
–The IP address of interface s0/2/0 (serial interface #0 on card #2 in module #0) is 10.200.1.5
–All IP addresses in range 10.200.1.1 – 10.200.1.254 are directly connected to interface s0/2/0 (that is, accessible through s0/2/0 without going through another router).

Front 11

Router Types
Customer internal (C) routers

Customer Edge (CE) router connects

Provider Edge (PE) router.


Back 11

Router Types
Customer internal (C) routers interconnect customer subnets at one site.
Customer Edge (CE) router connects customer site across an Access line to an ISP Provider Edge (PE) router.
PE routers interconnected through multiple Provider (P) routers within the ISP backbone.

Front 12

When and why was Cisco Express Forwarding developed

Back 12

Developed by Cisco in the 90's to make routers faster

Front 13

What is the slowness mechanism for performing routing?

Back 13

Process Switching
Router CPU handles packet forwarding

Front 14

What was the improvement upon Process Switching?

Back 14

Interrupt Switching
Interrupt driver handles forwarding

Where rather than a central CPU lower level interupt devices managed routing

Front 15

What router technology was purely based in hardware?

Back 15

Application-Specific Integrated Circuit (ASIC) Switching
Hardware ASIC handles forwarding

Front 16

What was needed for both Interrupt Switching and ASIC

Back 16

To use Interrupt or ASIC forwarding, a Forwarding Table must be built using Fast Switching or Cisco Express Forwarding (CEF)

Front 17

What complicates and causes delays in Process Switching?

Back 17

Process Switching
In Process Switching, an incoming packet is moved into CPU memory and Router CPU does processing:
Routing Table lookup to get Next Hop address
(recursive lookups to get connected address)
ARP Table lookup to get corresponding Layer 2 MAC address for next hop address
TTL field decrement
Recalculation of IP Header CRC
Fragment, if necessary
Create layer 2 header, pass to Outgoing Interface
This is slow and CPU can become bottleneck.

Front 18

How was CEF simplified

Back 18

CEF Table
To do Interrupt or ASIC Switching, a Forwarding Information Base (FIB) (also called CEF Table) is built that just contains essential information to forward packet.
Two methods to build FIB:
Fast Switching builds FIB on-demand when first packet of a flow arrives
CEF builds entire FIB in advance based on Routing Table. Forwarding Table may contain MPLS label that should be added to packet.

Front 19

Why is CEF needed for MPLS?

Back 19

CEF needed for MPLS
Router must have CEF enabled in order to use MPLS. This is done with ip cef global configuration command.
CEF builds two tables:
CEF Table (FIB), containing:
IP Prefix, Next Hop, Interface, MPLS Label(s)
Adjacency Table, containing:
Complete Layer 2 header for each Next Hop

Front 20

How is CEF Table different from Route Table?

Back 20

How is CEF Table different from Route Table?
CEF Table explicitly lists all known IP prefixes, broadcast addresses, multicast groups, etc.
CEF Table “Next Hop” values are always a directly connected neighbor of router. (No recursive lookups)
CEF Table may include MPLS label(s) that will be pushed onto packet if prefix matches.
CEF Table has no Distance measures to determine how “good” a route is. CEF Table is updated any time the Routing Table is updated.
CEF Table points to an Adjacency Table entry for each Next Hop.

Front 21

What are the two types of CEF load balancing?

Back 21

Two types of load balancing:
Per-packet: Packets sent over multiple paths in round-robin fashion
Per-destination: Packets hashed on <source IP, dest IP> onto 16 hash buckets - all packets in each hash bucket sent over same path.

Front 22

What configuration parameter is used to specify the number of hops?

Back 22

maximum-paths specifies how many next hops (paths) are allowed per prefix in the routing table.

Front 23

What is the problem with per packet load balancing?

Back 23

Packets may get out of order

Front 24

How do distance vector routing protocols perform the routing function?

Back 24

Distance Vector Protocols
•In Distance-Vector routing protocols, each router essentially sends a copy of its routing table to neighbor routers at regular intervals.
•Each neighbor computes its own distance metric for each destination subnet (routing prefix) and keeps the smallest one (shortest distance) in its own table.
•Examples:
–Routing Information Protocol (RIP) uses hop count metric.
–Cisco (Enhanced) Interior Gateway Routing Protocol (EIGRP, IGRP) uses bandwidth, delay, reliability, load, MTU and hop count as metrics.

Front 25

How Link State routing protocols perform routing function?

Back 25

Dynamic
•Each router shares knowledge about its connected links: link states (who I am directly connected to) and link metrics (link bandwidth, delay, reliability, load, other cost, hop counts etc.)
•Link state packets shared with every other router via multicast
•Link State Update packet sent out whenever there is a change
•OSPF (Open Shortest Path First), is an example.
•Encapsulated by IP
•General steps
–Send Hellos to discover reachability
–Download initial link database from neighbor
–Build link state packets (advertisements)
–Broadcast the link state packets when there are changes
–Build a map from the received link state packets
–From the map calculate the shortest path

Front 26

Distance Vector Protocols features

Back 26

•Dynamic
•Share knowledge about the entire AS
•Share only with neighbors
•Share at regular intervals
•Share the distance vector: a router’s distance to all other routers within the AS
•Routes are calculated based on the distance vectors received from the neighbors.
•It is based on the Bellman-Ford algorithm. (Don’t worry about the details of the algorithm)
•RIP (Routing Information Protocol) – the older Internet routing protocol, is an example
–encapsulated in UDP: Well known port 520.(
–RFC 1058 (RIPv1); RFC 1388 (RIPv2), RFC 2080 (RIPng for IPv6)

Front 27

Why use OSPF?

Back 27

RIP does not scale well
•OSPF provides load balancing
•OSPF provides partitioning of the network and forces you to develop an hierarchical design.
•OSPF provides for packet update authentication.
•OSPF multicasts rather than broadcasts.
•Support virtual links, an abstraction away from physical connections
•Support VLSM (variable length subnet masks)
•Provides for route summarization techniques
•Converges quickly
•Scales up to very large networks (hierarchical areas)

Front 28

What happens at power up on OSPF configured router?

Back 28

When router first powers up, it will
–Determine who its neighbor routers are.
–Download a complete Link State Database from a neighbor router

Front 29

What happens when something changes in OSPF?

Back 29

When something changes in the network (link goes up or down, cost metric changes, etc.)
–Discovering router will flood Link State Advertisements (LSAs) to all other routers in the Area to update their databases

Front 30

If a router shares a link with another router, then these two routers are considered ____________.

Back 30

If a router shares a link with another router, then these two routers are considered neighbors.

Front 31

An OSPF router will periodically measure the __.

Back 31

An OSPF router will periodically measure the metric (bandwidth, delay, throughput, etc.) between itself and each of its neighbors.

Front 32

What is important about OSPF (and MPLS) Router ID

Back 32

A single IP address ID identifies a router in OSPF (also in MPLS):
–Must be unique
–By default: ID is the highest IP address on an active interface. … That means it can change!
–Best idea: Configuring a loopback interface overrides the default. Loopback0 interface if present will always be router ID.
–You should always assign an IP to Loopback0 on all routers that use OSPF and/or MPLS!