Title: More on the IP
1More on the IP
2Internet Layer Process
- Transport layer process passes EACH TCP segment
to the internet layer process for delivery
Transport LayerProcess
TCP segment
Internet LayerProcess
3IP Connectionless Service
- The Internet Protocol (IP)
- Internet layer protocol
- IP messages are called IP packets
- No connections are established
- No open, close, error correction, flow control
- Low overhead
IP Packet
Internet Process
Internet Process
4IP Connectionless Service
- IP is unreliable
- No error handling (Let TCP catch errors!)
- No sequence numbers, so no way to put arriving IP
packets in order (Let TCP put the TCP segments
these IP packets contain in order!)
IP Packet
Internet Process
Internet Process
5TCP/IP Partnership
- TCP checks for errors once, at the destination
host - IP is used in many hops between routers
- Not checking for errors at each step greatly
reduces overall processing work - Reduces router costs
Check Only Once
Transport
Transport
Internet
Internet
Internet
Host
Router
Host
6Cobnnectionless IP
- IP is unreliable (does not catch errors)
- But this is not bad
- First, errors are caught--at the next-higher
layer (transport) if TCP is used - Second, avoiding error checking at each hop
between routers lowers router costs - Far less expensive to check for errors on one
destination host than on many routers along the
way
7IP is a Best-Effort Service
- IP Only Offers Best-Effort Service
- Does its best to get packets through
- No guarantees of delivery
- No way to give priority to time-sensitive
traffic, such as voice - Overall, low overhead but limited Quality of
Service (QoS) - QoS should change in the future (see Chapter 8)
8IP Address
- 32-bit Strings
- Often given in dotted decimal notation
128.171.17.13 - Fits into 32-bit source and destination address
field of IP headers
IP Packet
32-bit Source and Destination Addresses
9IP Addresses and Router Forwarding
- Routers use the destination IP address of an
incoming packet in the router forwarding
decision, that is, to decide what output port to
use to send the packet back out to the
destination host or to another router
B
B?
D?
Router A
D
Packet
C?
C
10IP Addresses
- Many Addressing Systems Use Hierarchical
Addressing - Postal delivery city, street address
- Post office looks at city first
- If not P.O.s city, put in bag for other city
- If in P.O.s city, put in bag for sorting by
street address - Hierarchical addressing greatly speeds sorting at
each post office - Imagine if we needed a sorting bin for each
address in the country!
11IP Addresses
- For IP, Routers Take the Place of Post Offices
- There are hundreds of millions of IP addresses on
the Internet - Routers cannot store decision rules for reaching
each address individually - So router simply asks if a destination IP address
is that of a host on one one of the networks or
subnets connected to the router or must be passed
on to another router - This is the router forwarding decision
12IP Addresses
- To Simplify Router Decisions, IP Addresses are
Hierarchical - The Internet is Made of Many Individual Networks
Owned by Different Organizations - First route packets to a single network only
need one sorting bin for each network! - In the next step, route packet to host on the
network
13IP Addresses
- Each Organization is Given a Network Part Number
- For the University of Hawaii, this is 128.171
- All IP Addresses in that Organizations Network
Begin with that Network Part
Network Part
IP Address
128.171
14IP Addresses
- Network Parts can be 8 to 24 bits long
- For University of Hawaii, it is 16 bits long
- 16 bits is only an example
Network Part (8 to 24 bits)
IP Address
15IP Addresses
- Between different organization networks, routers
look first at the Network Part of an arriving IP
packets destination address - If the network part is not that of the
organization, the router cannot deliver the IP
packet locally - Passes the IP packet on to another router, called
a next-hop router, to move the IP packet closer
to the destination host
Network Part
16IP Addresses
- Local Part
- The part of the IP address after the network part
is called the local part - Total address is 32 bits, so if the network part
is 8, the local part is 24
Network Part
Local Part
IP Address (32 bits total)
17Assigning Network Parts
- Organization applies to an Internet IP address
registrar - Registrar gives organization a network part
- Organization assigns the local part to its hosts
internally - Only large organizations and ISPs get network
parts
128.171.17.13
Registrar
Firm
128.171
128.171.123.130
18Assigning Parts
- Example
- IP address registrar gave the University of
Hawaii the network part 128.171 - UH gave the College of Business Administration
the subnet part 17 - College of Business Administration gave the host
part 13 to a computer it later gave the host name
voyager.cba.hawaii.edu - So the computers IP address became 128.171.17.13.
19IP Addresses
- Most Organizations Have Multiple Subnets within
the Organizational Network - Usually represent each as a Subnet Part within
the Local Part - Remaining Bits are the Host Part, designating a
particular host on that subnet
Local Part
Network Part
Subnet Part
Host Part
IP Address (32 bits total)
20Assigning Subnet Parts
- Organization Assigns Subnet Parts
- Assigns subnet parts to suborganizations
- Suborganization assigns host bits to hosts
128.171
128.171.17.13
Suborganization
Registrar
Firm
128.171.17.13
Host
21IP Addresses
- Within an organizational Network
- Router looks at Network Plus Subnet Part Combined
- If destination host is on a subnet attached to
the router, delivers the IP packet to the host - Otherwise, passes the packet on to a next-hop
router
Local Part
Network Part
Subnet Part
Host Part
IP Address (32 bits total)
22IP Addresses
- In IP Addresses, Network and Subnet are
Organizational Concepts, not Technical Concepts - Network is the collection of individual networks
and routers owned by an organization - Subnet is a collection of individual networks and
routers owned by a suborganization - Often a single physical network (subnet)
23Importance of Part Sizes
- Determine Number of Possible Networks, Subnets,
or Hosts - If There are N Bits in the Part, there can be 2N
possible Networks, Subnets, or Hosts - Actually, 2N-2
- All zeros cannot be used for a part
- All ones cannot be used for a part
- Example if part has 8 bits, 28-2 possibilities
(254)
24Masks
- IP Addresses are Always Paired with a Second
32-bit Number Called a Mask - Two Types Network Masks and Subnet Masks
- Network Mask Tells the Length of the Network Part
- Subnet Mask Tells the length of the Network Plus
Subnet Parts (not just subnet part) - IP Address will be paired with one or the other,
but not both simultaneously
25Masks
- Masks Begin with 1s, End with 0s (11100)
- For network masks, 1s are in Network Part bits
0s are in Subnet and Host Parts - For subnet masks, 1s are in Network and Subnet
Parts 0s are in Host part
11111111111111110000000000000000
26Masks
- IP Address-Mask Pairs often Written with Prefix
Notation - 128.171.17.13/16
- 16 means that the mask has 16 initial 1s
- Total number of bits is 32 in an IP address, so
there must be 16 trailing 0s
11111111111111110000000000000000
27Router Delivery
- If Destination Host is On the Source Hosts
Subnet, Source Host Delivers the Packet Directly - No router is involved
Source Host
Subnet
Subnet
Destination Host
28Router Delivery
- If Destination Host is NOT On the Source Hosts
Subnet, Source Host Sends the Packet to a Router
for Delivery
Subnet
Subnet
29Router Delivery
- If Destination Host is On One of the Routers
Subnets, the Router Sends the Packet to the
Destination Host for Delivery
Subnet
Subnet
30Router Delivery
- If Destination Host is NOT On One of the Routers
Subnets, the Router Sends the Packet to a
Next-Hop Router for Delivery - May have to choose among several possible
next-hop routers for delivery
Subnet
Subnet
31Router Delivery
- Border Routers Connect Networks, Not Subnets
- Select between next-hop router on own network or
on another network
Own Network
Other Network
32Router Forwarding Tables
- Allow Routers to Decide Whether Local Delivery is
Possible to Destination Host - Allow Routers to Select Next-Hop Router if Local
Delivery is Not Possible
33Router Forwarding Tables
- Router Compares Destination IP Address to Each
Row in Router Forwarding Table - If matches IP address, delivers according to
Delivery rule - So if destination address of IP packet is
128.171.17.13, router delivers packet locally
34Router Forwarding Tables
- Also Has a Mask Column
- Masks destination IP Address of packet
- If Mask in a row is 24 bits long, router only
compares first 24 bits of packets IP dest. addr. - Compares to IP Address Part for that row
35Router Forwarding Tables
- Also Has a Mask Column
- A network mask for a host outside the
organizations network - A subnet mask for an internal host
- Cant tell which by looking at the mask
36Router Forwarding Tables
- Also Has a Mask Column
- Masks destination IP Address of packet
- IP Address Part really is a network part or
network plus subnet part of a network or subnet,
respectively
Not in the Book
37Router Forwarding Tables
- Matching Destination IP Addresses
- Example Destination IP Address is 127.171.17.13
- Mask is 24, so only look at 127.171.17
- Matches rows IP address part, so use Local
delivery
38Router Forwarding Tables
- Longest Match Principle
- Must select one row to determine delivery
- If two rows match, use longest match, that is
match to greatest number of bits - For 127.171.17.13, use local delivery (24-bit
match)
39Router Forwarding Tables
- Metric
- If same length of match, turn to metric column
- Metric describes the desirability of a choice
- If metric is cost, choose lowest cost
- For other metrics (speed, etc.), may chose
largest value
40Router Forwarding Tables
- There May be No Matches
- One IP Address Part is Always 0.0.0.0
- If there is no match, choose its next-hop router
(called the Default Router)
41Router Forwarding Tables
- Recap of Selection Rules
- Compare destination IP address of an arriving
packet against ALL rows within the router
forwarding table because there may be multiple
matches - Select the single row that matches
- If multiple rows match, select the longest match
- If multiple rows tie on the longest match, select
the row with the largest or smallest metric,
depending on the specific metric - If there is no match, select the default router
row
42Router Forwarding Tables
- Delivery
- Table not only designates local delivery or a
next hop router - Also designates the router interface (port) that
will be used for delivery
43Dynamic Routing Protocols
- How Do Routers Get Information for their Router
Forwarding Tables? - Share router forwarding table information
- Standards for these exchanges are called dynamic
routing protocols
Router Forwarding Table Information
44Dynamic Routing Protocols
- How Do Routers Get Information for their Router
Forwarding Tables? - Thanks to dynamic routing protocols, the Internet
needs no central point of control - Routers create their router forwarding tables
strictly by information from peers and their own
knowledge
Router Forwarding Table Information
45IP Version 6
- Current Version of IP is IP Version 4
- This is the version we have been discussing
- Has 32-bit IP address fields
- Not long enough running out of IP addresses
- Next Version will be IP Version 6
- Will have 128-bit IP address fields
- Will allow vast numbers of IP addresses (2128)
- Being adopted slowly
46Terminology Confusion
- TCP/IP is a Standards Architecture
- Includes not only TCP and IP but also UDP, HTTP,
and many other protocols - May not even use TCP (UDP instead) or IP (ARP
instead, as discussed in Module A) - TCP and IP are Individual Standards
- Within the TCP/IP Architecture