A)
Introduction (267 – 294)
a.
Metropolitan Area Network (MAN) – a network that
expands into a metropolitan area and exhibits high data rates, high
reliability, and low data loss
a.
Wide area network – a network that expands
beyond a metropolitan area
B)
Metropolitan Area Network Basics
a.
Characteristics of a MAN Network
a.
Failover
– ability to reroute in the event of a failure. High redundant circuits, so
that in the event of a component failure, the network can quickly reroute
traffic away from the failed component
1.
Failover
time – speed at which a failover is performed
b.
Topology is based on a ring, logically and
physically, network routers and switches are interconnected in a ring fashion
c.
Ability that neither LANs or WANs currently
have, is the ability of a user to dynamically allocate more bandwidth on demand
b.
SONET versus Ethernet
a.
Network topology is a ring that is composed of
multiple rings that enable the network to provide backup in the event of a
segment failure
b.
Disadvantages of SONET
1.
Complex, fairly expensive technology that cannot
be provisioned dynamically
a.
Designed to support multiple streams of voice
channels and thus does not scale nicely
c.
Ethernet MAN
1.
Metro
Ethernet – data transfer service that can connect your business to another
business (or businesses) using a standard Ethernet connection
a.
Uses point-to-point connection
b.
Bandwidth
profile – describes various characteristics about the connection, such as
basic data transfer rates, basic burst rates (a surge of data that is
transmitted for a short period of time), excess data transfer rates, and excess
burst rates
C)
Wide Area Network Basics
a.
Wide area
network (WAN) – a collection of computers and computer-related equipment
interconnected to perform a given function or functions, typically using local
and long-distance telecommunications systems
a.
Station
– a device that a user interacts with to access a network, and it contains the
software application that allows someone to use the network for a particular
purpose
b.
Node
– a device that allows one or more stations to access the physical network, and
is a transfer point for passing information through the network
1.
Subnetwork
(network cloud) – a collection of nodes and interconnecting
telecommunications links
b.
Types of network clouds
a.
Circuit-switched
network – a network cloud in which a dedicated circuit is established
between the sender and receiver, and all data passes over this circuit
b.
Packet-switched
network – all data messages are transmitted using fixed-sized packages,
called packets, and no unique, dedicated physical path is established to
transmit the data packets across the subnetwork
1.
Datagram –
packet-switched network, each data packet can follow its own, possibly unique,
course through the cloud
2.
Virtual
circuit packet-switched network – all packets that belong to a logical
connection can follow the same path through the network
c.
Broadcast
Network – transmits its data, the data is received by all other nodes
c.
Connection-oriented versus connectionless
network applications
a.
Connection-oriented
network application - provide some guarantee that information traveling
through the network will not be lost and that the information packets will be
delivered to the intended receiver in the same order in which they were
transmitted
1.
Reliable
service – network that requires that a logical connection be established
between two endpoints
b.
Connectionless
network application – does not require a logical connection to be made
before the transfer of data. Thus, does not guarantee the delivery of any
information or data. Data may be lost, delayed, or even duplicated
D)
Routing
a.
Weighted
network graph – network graph in which each pair of nodes can be assigned a
weight or associated cost
b.
Dijkstra’s
least-cost algorithm – calculates a least-cost path through a network.
Executed by each node and results are stored at the node and sometimes shared
with the other nodes
c.
Flooding
– each nodes takes the incoming packet and retransmits it onto every outgoing
link
a.
Rules for flooding
1.
A node need no send a copy of the packet back to
the link from which the packet just arrived
2.
Hop limit
(network limit) – can be placed on how many times any packet is copied. Counter
is called the hop count
d.
Centralized versus distributed routing
a.
Centralized
routing – involves storing all the routing information at one central
location. Rarely used in wide area networks
b.
Distributed
routing – uses a routing algorithm, such as a least-cost algorithm, to
generate routing information and dictates that this information be stored at
distributed locations – typically, routers – within the network
1.
No single node (or central router) is
responsible for maintaining all routing information
a.
If any node crashes, it will probably not
disable the entire network
b.
A node will not need to send a request to a
central router because each node has its own table
e.
Adaptive versus fixed routing
a.
Adaptive
routing – a dynamic technique in which routing tables react to network
fluctuations, such as congestion and node/link failure
b.
Fixed
routing – routing tables are created once, typically when the network is
installed, and then never updated again
f.
Routing examples
a.
ARPANET – distance vector routing algorithm, an
adaptive algorithm in which each node maintained a routing table called a
vector, the routing algorithm was also a distributed algorithm. Every 30
seconds, each node exchanged its vector with its neighbor. Also called Router information protocol (RIP)
b.
Link state routing algorithm
1.
Steps
a.
Measure the delay or cost to each neighboring
router
b.
Construct a link state packet containing all of
this timing information
c.
Distribute the link state packets via flooding
d.
Compute new routes based on the updated
i.
Once a router collects a full set of link state
packets from its neighbors, it creates a routing table, usually using
Dijkstra’s least-cost algorithm
ii.
Open
shorted path first (OSPF) protocol – a link state algorithm that is still
used today by many internet routers
E)
Network
congestion – when a network or a part of a network becomes so saturated
with data packets that packet transfer is noticeably impeded
a.
Possible solutions to congestion
a.
Implicit
congestion control – application is simply observing its own throughput and
not relying on any special types of signals coming from the network
b.
Explicit
congestion control – when the network signals the transmitting station to
slow down
c.
Forward
explicit congestion notification (FECN) – when a frame relay router
experiences congestion, it sends a congestion signal (inside the data frames)
forward to the destination station, which in turn tells the originating station
to slow down the transfer of data
d.
Backward
explicit congestion notification (BECN) – the frame relay router
experiencing congestion sends a signal back to the originating station, which
then slows down its transmission
e.
Buffer
preallocation – before one node sends a series of n packets to another node, the sending node inquires in advance
whether the receiving node has enough buffer space for the n packets. If the receiving node has enough buffer space, it sets
aside the n buffers and informs the
sending node to begin transmission
f.
Connection
admission control – used in Asynchronous transfer mode (ATM), avoid
congestion by requiring users to negotiate with the network regarding how much
traffic they will be sending, or what resources the network must provide to
satisfy the user’s needs before the user sends any data. If network cannot
satisfy the user’s demands, the user connection is denied
g.
Service
level agreement – a legally binding, written document that can include
service parameters offered in the service, various types of service/support
options, incentives if the service levels are exceeded, and penalties if
service levels are not met
No comments:
Post a Comment