CS 164: Global Internet

1

• CS 164 Global Internet
• Slide Set -- 11

2
In this set ...

• More about subnets
• Classless Inter Domain Routing (CIDR)
• Border Gateway Protocol (BGP)
• Areas with OSPF

3
Forwarding on Subnets

• Resolution of subnet address Bitwise ANDing Host
  IP address with Subnet Mask gives subnet number.
• When a host wants to send an IP packet
• Perform BITwise AND between subnet mask and
  destination IP address
• If result its subnet no. destination is on
  same subnet (Send ARP etc.).
• If not, send packet to default router R.

4
Router Forwarding Tables

• Table holds entries like ltSubnetNumber,
  SubnetMask, NextHopgt
• Router ANDs dest addr with subnet mask of each
  entry.
• Find the right entry (Match with subnet no.) and
  forward to Next hop.

28
28
Subnet Number Subnet Mask Next Hop
128.96.34.0 255.255.255.128 Int 0
128.96.34.128 255.255.255.128 Int 1
128.96.33.0 255.255.255.0 R2
Router R1s table
5
Other Issues

• Subnet Mask need not align with byte boundaries
  (e.g. 255.255.255.128) -- 7 zeroes.
• Non contiguous masks are possible -- 255.255.1.0
  -- however, this makes administration difficult
  -- not recommended.
• One could have multiple subnets on the same
  physical network ! However, now, hosts on the
  same net would need to go through a router in
  order to talk to each other.

6
Outside View

• Routers outside a group of subnets see the group
  as a single network -- e.g. 128.96
• However, once packet arrives to the group,
  routers within the group need to forward the
  packets to the proper subnet.

7
Classless Interdomain Routing

• Abbreviated as CIDR.
• If a network grows to more than 255 hosts, it may
  want a Class B address.
• One possible way of avoiding is to handle many
  Class C routing addresses -- but then, for this
  one network, each router has to maintain multiple
  routing entries.
• CIDR is an attempt to balance the desire to
  minimize the number of routes that a router needs
  to know versus the need to hand out addresses
  efficiently.
• Key property CIDR enables aggregation of routes !

8
Removing rigid boundaries

• The idea is to break the rigid boundaries between
  classes.
• As an example, if a network grows to about 16x255
  hosts, assign a contiguous block of Class C
  addresses as opposed to a Class B address.
• Example 192.4.16 to 192.4.31
• Note -- 16 Class C addresses better than 1 Class
  B in terms of address efficiency.
• In the above example -- the top 20 bits are the
  same for all the addresses and so we have
  effectively created a 20 bit network number !

9
The Prefix

• The 20 bit address in our previous example is
  called the common prefix for the set of
  addresses that are allocated.
• Observe -- what we did was that we allocated a
  block of Class C addresses that shared a common
  prefix.
• Now, with this new representation, the network
  numbers are represented by ltlength,valuegt -- the
  length represents the number of bits in the
  prefix.

10
Subnets vs CIDR

• The concept is similar but
• In a subnet, a single address is shared among
  multiple physical networks.
• With CIDR, we collapse multiple network addresses
  into a longer network address that is typically
  assigned to an AS (the single AS would have a
  network number or prefix that reflects the block
  of addresses).
• Thus, when we want to route to any of the
  networks or even subnets within the AS, we route
  to the AS.

11
Route Aggregation

• Specifying simply the prefix associated with an
  AS (as opposed to stating the subnet number
  explicitly) is called route aggregation.
• When sending route advertisements (we will see
  how), it suffices to simply advertise common
  prefixes.
• Note that for this, careful planning would be
  needed.

12
An Example

• Border gateway advertises the common prefix only!

13
Longest Prefix Match

• Prefixes may overlap
• Example 171.69 and 171.69.10 may be found in the
  forwarding table of a single router.
• Now, if the destination is 179.69.10.7, both the
  prefixes match !
• Policy -- Choose the longest prefix. why ?
• Choosing the longest prefix the right choice
  since an organization may switch ISPs.

ISP 1 223.1.1.0
ISP 2
ISP 2 would advertise 223.1.1.240 explicitly
223.1.1.240
14
Revisiting Autonomous Systems

• ASes provide an additional way of hierarchically
  aggregating routing information in the Internet.
• AS --gt also called domains and can run their own
  protocols within their administrative regimes.
• In each AS, the amount of routing information may
  be dramatically reduced by using a default
  router.
• If the corporate network is connected to the
  Internet by means of a single border router nodes
  simply send messages to this router.

15
Internet View Revisited
Multihomed -- no transit traffic.
Stub AS -- only local traffic
Service Provider Networks
16
Some notation

• Peering Point Points where different providers
  interconnect.
• Local Traffic Traffic that begins and terminates
  on nodes within an AS.
• Transit traffic -- passes through various Ases --
  backbones carry transit traffic.

17
Exterior Gateway Protocol

• Abbreviated as EGP
• Hierarchical routing
• Reach higher level in the hierarchy
• Tree structure for routing topology.
• No peer-to-peer communications.

18
Border Gateway Protocol

• Abbreviated BGP -- currently version 4 and is in
  use.
• Rather complex.
• Goals of BGP are modest
• Any loop free path is to be found between the
  source and destination (not necessarily min cost
  or shortest).

19
Why the modest goal ?

• Each AS may have a different set of nodes, so it
  is unclear which route is the min-cost route!
• Route aggregation also difficult
• Lots of routing information is required in order
  to guarantee optimality -- may be infeasible.
• Trust -- misconfiguration may not yield optimal.

20
BGP Details

• BGP supports flexibility -- paths could be chosen
  by a provider based on a policy.
• To configure BGP, each AS admin picks at least
  one node to be the BGP speaker -- a
  spokesperson node for the entire AS.
• The BGP speaker establishes a BGP session with
  other BGP speakers in other ASes.
• In addition, there are border gateways using
  which packets enter/leave ASes.
• Source advertises complete paths (unlike distance
  vector or link state routing) -- thus loops are
  prevented.

21
An Example

• AS 2 says 128.96, 192.4.15, 192.4.32, 192.4.3 can
  be reached via AS 2.
• AS 1 advertises that these networks can be
  reached via ltAS1, AS2gt --note full path
  description.
• Loops are avoided.

22
AS Numbers

• Each AS is assigned a number -- 16 bits and is
  unique.
• The uniqueness requirement has been relaxed to
  some extent --stub ASes do not need to be unique.
• One could have up to 65 K AS numbers.

23
BGP Messages

• BGP has four types of messages
• OPEN Establish a connection with a BGP peer
• Note BGP connection is TCP based ! (Port no.
  179).
• UPDATE -- advertise or withdraw routes to a
  destination
• Note --BGP speaker needs to be able to cancel
  previously advertised paths if nodes or links
  fail. This form of negative advertisements are
  said to advertise withdrawn routes.

24
BGP Messages (cont)

• KEEPALIVE Inform a peer that the sender is still
  alive but has no information to send.
• NOTIFICATION Notify that errors are detected.

25
BGP Message Format

• 16 byte fields.
• For more detail look at book.
• Important thing --- BGP updates are of the type
  prefix/length
• 192.4.16/20
• Note that forwarding entries can also be
  similarly represented.

26
Routing with BGP

• For stub AS -- border router injects a default
  route into the intra-domain routing protocol.
• If there are more than one border router, each
  injects specific routes that they have learned
  from outside the AS.
• IBGP or Interior BGP is used to distribute the
  information to all other routers in the domain
  (and the speaker).

27
Routing Areas

• Especially used with OSPF.
• Subdomains of larger domains.
• One special area called backbone area. (Area 0).
• Within each area -- link state routing.
• Link state advertisements of non border routers
  do not leave area.
• Packet goes from non-backbone area to backbone
  area and crosses the backbone into the Internet.

• A router that is a member of both the backbone
  and a non-backbone area (R1) is called a area
  router.

28
Areas (continued)

• Border routers summarize routing information
  and make it available to other areas -- act like
  proxies --reflect costs to reach networks from an
  area.
• When there are many possible routes, routers
  choose cost info to forward packets.
• Trade-offs -- Optimality versus scalability --
  All packet have to pass through the backbone area
  (may not be optimal).

29
Next Time

• IPv6
• Introduction to the transport layer.