Chapter7

advertisement
Chapter 7
Local Area Network
Communications
Protocols
The Network Layer


Local area network communications protocols can be
thought of as the language of computer networks: They
provide the language and grammatical rules that define
communication
This chapter focuses on local area network protocols,
which control communication between hosts on interconnected networks
LAYER THREE: THE NETWORK
LAYER




The third layer of the OSI Network Reference Model is
the network layer
The network layer is primarily concerned with providing a
means for hosts to communicate with other hosts on
different network segments.
As explained in chapter 4, the datalink layer provides a
means for two hosts on a common network segment to
communicate.
Technologies such as Ethernet and token ring provide
this intra-segment connectivity.



Network layer protocols expand the capabilities of the
network by providing a means of delivering data (called
packets) between network segments.
A network layer protocol provides a means of addressing
a host on the interconnected network and a means of
delivering data across the network to destination hosts.
Addressing is the process of defining where on the
internetwork the host is located and routing is the
process of determining the best path to the destination
host and delivering the data




The datalink layer (layer two) transmits frames of data.
The network layer (layer three) transmits packets of data
There are two key levels of networking: single segments
where every host receives every packet and multisegment networks where some internetworking device
forwards packets between network segments
Single network segments are commonly referred to as
segments, sub-networks, or subnets.
Inter-connected segments are commonly referred to as
networks or internetworks.


Don’t let these varying terms confuse you: if every host
sees every frame of data on the LAN, it’s a segment.
If more than one segment is used, it’s an internetwork
Network Layer Addressing




The role of network layer addresses is to provide a
means to uniquely identify a host on the internetwork.
The network layer address is used in the routing process
to deliver a packet of data to the correct network
segment for delivery.
In this manner, network addresses are used for “end-toend” or “inter-segment” communication
There are two basic components to a network layer
address: a network segment address and a host address.



The network segment address identifies in which network
segment the destination host is located.
This address is used by the routing process to determine
the destination network segment and deliver the packet
to that segment.
Once the packet has arrived at the correct network
segment, the host address is resolved to a physical
address and passed to the datalink layer for delivery




Collectively, these two parts identify a host on the
internetwork.
The segment address must be unique to the internetwork
to ensure that the packet is delivered to the correct
network segment.
Similarly, the host address must be unique within a
network segment.
As shown in Figure 7-2, it is possible to use the same
host address on multiple segments.




There is no chance for confusion between the hosts, as
the packet will be routed to the network segment of the
correct host based on the segment address.
Network layer addresses are assigned on a per-NIC basis.
It is possible for a single device to contain more than one
NIC.
In this case, each NIC will have its own network layer
address (segment address + host address).
Devices that contain more than one NIC are said to be
multi-homed
Network Layer Addressing


Figure 7.2: The host address must be unique within
a network segment.
The packet will be routed to the network segment
of the correct host based on the segment address



There are several reasons to multi-home a device,
including performance, reliability, and stability.
However, the primary reason a device is multi-homed is
to allow it to forward packets from one network segment
to another.
By definition a device that performs packet forwarding is
known as a router
Network Layer vs. Datalink Layer
Addressing


The network layer is primarily responsible for addressing
hosts uniquely on an internetwork and providing a means
of delivering data across network segments.
The responsibility of moving data within a network
segment is that of the datalink layer uses the physical
address (also known as the MAC addresses for Ethernet
and token ring technologies) of the NIC to deliver data
rather than the network layer host address datalink layer



The process of determining the physical layer address of
an NIC from the network layer address is known as
address resolution
Each NIC is assigned a single data link layer physical
address and one or more network addresses.
The network layer address is used by upper-layer
protocols to denote which host they wish to communicate
with





The process of delivering data to a destination host is
analogous to the method used to deliver a letter to a
person’s house.
The street name (network segment address) is used to
route the mail to the correct postal route.
The mail carrier for that route then looks at the house
number (host address) to determine the final destination
of the letter.
The mail carrier then resolves the house number to a
physical mail-box location (physical address) for delivery.
This relationship is shown in Figure 7-3
Protocol Encapsulation/Deencapsulation

Layer three packet will always contain at least three
sections:
 Source network address,
 destination network address, and
 Data




As illustrated in Figure 7-4, a packet of data from the
network layer is placed in the data section of a datalink
layer frame.
Known as encapsulation, this process adds both a header
and trailer to the packet.
The bit stream is finally passed along the shared media
that connects the two computing devices.
When the full bit stream arrives at the destination server,
the reverse process of encapsulation, de-encapsulation
takes place
Protocol Encapsulation


Figure 7.4: A packet of data from the network
layer is placed in the data section of a
datalink layer frame.
This is encapsulation
Fragmentation




The maximum size of a packet or frame of data is known
as the Maximum Transmission Unit (MTU)
In the event that a higher-level packet will not fit into a
lower layer’s available payload area, the higher-level
packet is broken into two or more packet fragments.
These packet fragments are sent across the network and
re-assembled into a single packet before being sent back
up the protocol stack on the destination host.
Fragmentation allows large quantities of data to be sent
across the network in smaller, more manageable
“chunks” of data


Data fragmentation is required between the application
layer and the network layer for large quantities of
application layer data (such as a 10 MB file) to be sent
across a network
When the source creates a packet that is larger than the
underlying layer two transport protocols, each router on
path between the source and destination must break the
packet into multiple fragments, sending each fragment
across the layer two link to the next router.
Packet Fragmentation

Fragmentation allows large quantities of data
to be sent across the network in smaller, more
manageable “chunks” of data.



Each frame that contains a fragment of the packet must
also contain a copy of the packet’s header to ensure that
the packet is properly reassembled
The next router must collect the fragments, re-assemble
them into the original network layer packet, and repeat
the process for the next hop.
Such repeated packet fragmentation and reassembly
places a large processing burden on routers, effectively
reducing their overall routing capacity
Packet Fragmentation Eliminated


Repeated packet fragmentation and reassembly
places a large processing burden on routers,
effectively reducing their overall routing capacity.
Solution? Lower Layer 3 packet size!
Routing


Routing is the process of moving data across network
segments toward its final destination.
Routers receive frames of data, de-encapsulate the layer
three packet, examine the network layer packet header,
determine the next hop of the packet, package the
packet into a new data frame and transmit the new
frame
Routing Is Address Processing

Perhaps the most important thing to understand about
routing is that it is nothing more than address processing
performed when messages need to travel beyond the
local LAN
Routing as Address Processing



The default gateway or local router receives the datalink
layer frame explicitly addressed to it and examines the
ultimate destination address held in the packet.
The router then consults its routing tables to see if it has
an entry for a known path to the ultimate destination
workstation.
That known path may be via another router, or the
ultimate destination workstation may be part of a
different LAN connected to this same router through a
different NIC.

In either case, the packet and its addresses are not
modified but are instead re-encapsulated in a fresh
datalink layer frame with the physical layer destination
address of either the ultimate destination workstation, or
of the next router along the path to that workstation
Routing Tables




Hosts and routers decide where to send packets by
looking up the destination address in their routing table.
A routing table consists of a series of destination
networks, the address of the local router that provides
service to the destination network, and a cost associated
with the route.
The cost is used to determine the best route in the event
that there are multiple routes to the destination available
Regardless of protocol, there should always be a special
entry in the routing table for the default router or
gateway of last resort.

This is the router where a packet should be sent if there
is not a route listed in the local routing table for the
destination network
Routing Protocols



when a router is initially started, it only knows about the
interfaces connected to it, or static routes that have been
configured by an administrator.
In order for a network to dynamically build
comprehensive routing tables that automatically add new
routes and remove old ones, a routing protocol must be
used.
Routing protocols provide routers a means of
automatically exchanging routing tables to ensure that
each router knows where to route packets for a given
destination




There are two basic classes of routing protocols: interior
gateway protocols and exterior gateway protocols.
The difference between interior and exterior gateway
protocols is the scope of the routing information they
distribute.
Interior gateway protocols distribute routing information
within a hierarchical address space such as that assigned
to a single organization.
As shown in Figure 7-8, these hierarchical address
spaces, also known as autonomous units, can be
interconnected into an internetwork.


When autonomous units are connected into an
internetwork, exterior gateway protocols are used to
distribute information about the various autonomous
units between the routers that provide connectivity
between them
Just as the post office first needs to get a letter to the
right city before worrying about the street names and
house numbers, an internetwork must get a packet to
the autonomous unit that contains the destination
network segment and host
Routing Protocols




There are two major categories of interior routing
protocols: distance vector and link state.
Distance vector protocols broadcast their entire routing
table periodically.
In this manner, changes to the network routing tables
slowly make their way through the network.
A router using a distance vector algorithm knows nothing
about the make-up of the network beyond the next hop
to the destination




Link state protocols transmit a more complete picture of
the network between routers.
Through the use of link state packets (LSP), each router
learns the structure of the entire network.
In this manner, the link state algorithm can make better
routing decisions.
Link state routing reacts quicker to changes in the
routing structure than distance vector routing while using
less bandwidth maintaining routing tables
Download