A network manager

Download Report

Transcript A network manager

Arab Open University - AOU
T209
Information and Communication
Technologies: People and Interactions
Thirteenth Session
Prepared by: Eng. Ali H. Elaywe
1
Reference Material
 This session is based on the following
reference:


Book NM: Network Management
Book E: Experiments
Prepared by: Eng. Ali H. Elaywe
2
Topic 1: Introduction
 Local area networks (LANs) have become complex
systems. They comprise a myriad of manufacturers’
products linked in various ways and connected to other
networks, as illustrated in Figure 1
 An ever-increasing dependence on networked services
for critical business functions, such as placing orders with
suppliers, supplying organizational information to clients,
and providing reliable information to managers, means
that the role of the network manager has become vital
within the organization
Prepared by: Eng. Ali H. Elaywe
Continue
3
Figure 1 Network links in a typical organization
Prepared by: Eng. Ali H. Elaywe
Continue
4
 In Book NM, you will see how management and
organizational techniques, introduced in Book M, are
used to manage computer networks. The following topics
will be introduced here:



Network management structure
Management functions
Management tools
 Once you have completed Book NM, or earlier if you
wish, you may use the network simulation software to
build and test simulations of computer networks. Book E
gives you details of how to load and run this software,
along with exercises to enhance your studies of this
module
Prepared by: Eng. Ali H. Elaywe
5
Topic 2: Network Management
 Organizations need a well-managed
computer network to provide users with the
services essential for their day-to-day work
 Activity 1 (reflective)

Make a brief list of what you consider are the
characteristics of a well-managed computer
network, from the perspective of a user?
Continue
Prepared by: Eng. Ali H. Elaywe
6

As a user, I would expect a well-managed computer network to
have the following characteristics:
 1- Reliability – the service should be available on request
 2- Error-free – it should cause no changes to the information
during transmission
 3- Responsive – there should be a minimum operational delay in
providing a service
 4- Efficiently organised and improved – it should accommodate
new users, services, applications and technologies.
 5- Secure – it should offer only authorised access and a virus-free
network

Note: You might list others as well !!
Prepared by: Eng. Ali H. Elaywe
Continue
7
 The three essential constituents of the network
management structure, illustrated in Figure 2 overleaf are:



1- A network manager (a person)
2- A network management system (the management tools)
3- A network (managed network elements and
communication links)
 The network manager makes decisions based upon
information about the operation of the network. Operational
control will be exerted over the network, and action taken
when problems are reported
Prepared by: Eng. Ali H. Elaywe
Continue
8
Figure 2 A network management
structure
Prepared by: Eng. Ali H. Elaywe
9
Sub-Topic 2.1: The role of the network
manager
 The challenge to the network manager is to ensure
that the network infrastructure consistently
services existing requirements and yet will be
able to meet the demands from new ones
 This challenge, illustrated in Figure 3, has to be met
within the constraint of budgets, the need to meet
organizational objectives and the expectation of
users that access to the network will always be
available when they want to use it
Prepared by: Eng. Ali H. Elaywe
Continue
10
Figure 3 The network manager’s challenge
Prepared by: Eng. Ali H. Elaywe
Continue
11
 Managing computer networks is as much a ‘craft’
as it is a ‘science’


The ‘craft’ attributes include an understanding and
knowledge of people, relationship insights and an
organizational understanding
The ‘science’ involves the technical attributes of
computer networking that balance and reinforce the ‘craft’
 The complexity of the network will govern the
tools and level of technical understanding
necessary to ensure the best possible service to
the users
Prepared by: Eng. Ali H. Elaywe
Continue
12
 Project Management


One central task undertaken by the network
manager is that of project management
According to Kerzner (1998), a project is a
unique set of activities and tasks that consume
resources (people, money and equipment); it
has specific objectives which have to be met,
including defined start and end dates and
funding limits
Prepared by: Eng. Ali H. Elaywe
Continue
13
 The following phases of a project represent a client–
contractor model, which will serve well as a model of the
usual relationships between the network manager and
clients:





1- Requirement specification: a formal description of the needs of
the client, including time-scales and expected costs
2- Design: a detailed description of how the client’s requirements will
be met, including actual costs and time-scales
3- Implementation: undertaking the work as agreed in the design
4- Testing: ensuring the implementation actually meets the client’s
requirements
5- Maintenance: a continual monitoring of the implementation to
ensure it continues to meet the original requirements, with
rectification in the case of failure, and updating as necessary
Prepared by: Eng. Ali H. Elaywe
14
Sub-Topic 2.2: Network element
mangement
 Let’s now consider network elements that are capable of being
managed. By this, we mean that they are able to accept
management commands and respond in an appropriate manner.
Figure 4 shows examples of networked elements which can have
this capability
 Managed elements are remotely monitored and controlled over
the network by special network management application
software run on a dedicated network management workstation


Thus, managed elements must be capable of storing and communicating
information to, and receiving control from, the management application
software
Information gathered from using this software can help to solve problems
and plan future enhancements of the network. It can also be used with a
network simulator to provide an exact analogy of the network functions.
Network simulation is discussed in Sub-Topic 4.1
Prepared by: Eng. Ali H. Elaywe
15
Continue
Figure 4 Examples of managed elements
Prepared by: Eng. Ali H. Elaywe
Continue
16
 Activity 3 (self-assessment)

You will notice that repeaters do not appear in Figure 4.
By referring back to your study of Book N, try to think why
they do not have management capability

A management capability requires some processing ability to be
able to action commands and provide responses. These functions
are not available in simple elements such as repeaters.
Repeaters simply regenerate the signal and repeat the packets
on all ports. (They do not contain any components with the
capability to process management functions.)
Prepared by: Eng. Ali H. Elaywe
Continue
17
 Two important characteristics had to be observed by the
designers when implementing these functions into the
elements and onto the network:
 1- Firstly, the communication and control requires extra
processing to respond to such demands
 2- Secondly, monitoring and control usually takes place
over the network medium and consequently increases
the level of traffic, i.e. the number of packets on the
network
Prepared by: Eng. Ali H. Elaywe
Continue
18
 Therefore, the two overhead consequences of managing
the network are:


1- The processing overhead on the managed element, due to
the management functions, should not appreciably affect the
primary function of the element
2- The increase in traffic caused by management monitoring
and control should not significantly load the network being
managed
 So whilst it is necessary to have network information and to
remotely control the operation of network elements, the
performance of the network must not be significantly
degraded by the management functions
Prepared by: Eng. Ali H. Elaywe
19
Sub-Topic 2.3: Network management
system
 A network management system is shown in Figure 5. It consists
of the following components:






1- A dedicated workstation (a networked computer)
2- Network management software running on the workstation
3- Managed network elements as discussed in Sub-Topic 2.2
4- Software agents, small programs with dedicated tasks, running in
the network elements
5- Simple Network Management Protocol (SNMP), a standard
protocol that specifies how information is carried between the agent
and the network management software. SNMP is discussed in SubTopic 2.5
6- A communications path, typically the network itself, (although it is
possible to manage a network remotely using modem links)
Prepared by: Eng. Ali H. Elaywe
Continue
20
Figure 5 A network management system
Prepared by: Eng. Ali H. Elaywe
Continue
21
 Examples of network management
packages include:



D-Link’s D-View®
HP’s OpenView®
Network Associates’ Sniffer®
Prepared by: Eng. Ali H. Elaywe
22
Sub-Topic 2.4: Network management
software
 Network management software consists of a set of useful
applications. These will aid network configuration, faultfinding,
monitoring, planning and intrusion control. These features
are a subset of the OSI ‘Framework for Network Management’
discussed in Book M, and offer facilities to:




1- Configure the network topology by controlling the addresses of
the network elements, (e.g. the name, location and IP address)
2- Manage faults by partitioning off a suspect collision domain
whilst the fault is rectified
3- Monitor the network performance by accessing the managed
elements and displaying current statistics in a graphical format to
evaluate the behaviour and effectiveness of the network
4- Restrict unauthorised users’ access to the network resources
by application of intrusion control procedures
Continue
Prepared by: Eng. Ali H. Elaywe
23
 Along with these management activities, the
program may offer such features as:



1- Automatic discovery – allows the software to find
managed elements on the network. An example is
illustrated in Figure 6 where 14 managed elements
have been ‘discovered’
2- Connectivity testing – offers a simple method of
checking whether a network element is active
3- Graphical displays of performance analysis, the
status of network elements and site maps
Continue
Prepared by: Eng. Ali H. Elaywe
24
Figure 6 A result from running ‘automatic discovery’
Prepared by: Eng. Ali H. Elaywe
25
Sub-Topic 2.5: Simple Network
Management Protocol
 At the heart of any network management
software is the protocol necessary to
communicate with the managed network
elements. Simple Network Management
Protocol (SNMP) was developed by the
Internet Engineering Task Force

It has become a standard protocol for network
management activities
Prepared by: Eng. Ali H. Elaywe
Continue
26
 Simple Network Management Protocol (SNMP) is used to
carry information between the management software and an
agent within the managed element. Refer back to Figure 5 to
remind yourself where these two items are situated within the
network management system
 SNMP sits above the transport layer in the OSI Reference
Model, as shown in Figure 7. It uses User Datagram Protocol
(UDP) to transport the messages

UDP has a simple packet structure offering a minimal overhead to
the network
Prepared by: Eng. Ali H. Elaywe
Continue
27
Figure 7 SNMP situated in the OSI Reference Model
Prepared by: Eng. Ali H. Elaywe
Continue
28
 SNMP only uses three commands:



1- Get-something: which causes a response from the managed
element
2- Set-something: this changes a feature of the managed
element
3- Trap-something: the managed element reports a problem to
the manager
 SNMP (the manager) requests the managed network
element (the subordinate) to either find something out
or do something
 Note that the element themselves only reports
autonomously when things go wrong
Prepared by: Eng. Ali H. Elaywe
Continue
29
 1- Get-something command:

A- Get-information command



On receipt of a Get-something command from the manager,
the software agent within the network element interrogates its
own database, called a management information base
(MIB), to obtain the requested information
An example of a response to a Get-information command is
shown in Figure 8
The information returned from this managed element
includes a description of the element, how long it has been
active and details of where it is situated
Prepared by: Eng. Ali H. Elaywe
Continue
30
Figure 8 A tabulated response to a Get-information command
Prepared by: Eng. Ali H. Elaywe
Continue
31

B- Get statistics command


Information about the types of packets, size of packets and
error packets that have passed through the element could be
requested
An example of statistical information is shown in Figure 9.
Many different statistics may be asked for, depending upon
the function of the element. I have chosen a graphical
display rather than just the tabulated values
 In general, a Get command will return a coded
numeric value (i.e. data), which must be
interpreted by the network management
application. Meaningful information will then be
displayed in the most appropriate format
Prepared by: Eng. Ali H. Elaywe
Continue
32
Figure 9 A graphical response to a Get-statistics command
Prepared by: Eng. Ali H. Elaywe
Continue
33
 2- Set-something command:





A Set command provides a means of remotely controlling
facilities in the network element
You may have, at some time, changed a facility on a modem or
printer
Certain characteristics may have to be changed physically, for
example by closing a switch
On a network, it can be very time-consuming to physically alter
settings on an element, especially if the element is at another site
The Set command allows the network manager to change element
settings remotely. When using vendor-specific software a graphical
display of the element will enable the manager to visualise what is
happening. An example of such a display is illustrated in Figure 10
Prepared by: Eng. Ali H. Elaywe
Continue
34
Figure 10 A managed element
Prepared by: Eng. Ali H. Elaywe
Continue
35
 3- Trap-something command:



A Trap command causes an unsolicited
communication to the management program from a
network element when an abnormal occurrence has
been detected
Typically, this would be when a certain number of lost
packets has been recorded or too many collisions
have occurred
The Trap command would be initiated because of a
pre-set threshold limit being exceeded. The network
manager using a Set command would specify this
limit
Prepared by: Eng. Ali H. Elaywe
Continue
36
 Protocol Data Units (PDUs)

The three SNMP commands, Get, Set and Trap are known as
protocol data units (PDUs). In detail they are:





A- GetRequest: asks for the value of a variable stored in the
management information base (MIB) of the element
B- GetNextRequest: requests information on multiple variables from
the MIB and is used to reduce network traffic. If one or more of the
requested variables is unavailable, no values are returned and an
error is reported
C- GetResponse: is sent by an agent in reply to one of the above PDUs
as a confirmation response along with the requested data
D- SetRequest: sets a MIB variable to a desired value
E- Trap: an unsolicited message sent from an agent to the manager,
usually to notify an alarm or similar event occurrence
 Note: whilst simple in operation, SNMP is a very complex
protocol to describe
Prepared by: Eng. Ali H. Elaywe
Continue
37
 Activity 4 (self-assessment)

Can you think of a major problem that might arise, as a
result of using the network to carry the management
information, should a serious fault occur on the network?

If the network fails you cannot get any network management
information in order to find out what has failed. The
management information is carried over the same network as
the organization’s information. This might be considered a
weakness of the system
Prepared by: Eng. Ali H. Elaywe
Continue
38
 Figure 11 shows the relationship between the
components within the network management
system, and are summarised as follows:




The management information base (MIB) is a resident database
in the managed element that holds management information about
the element
The agent running in managed elements on the network responds to
the PDUs from the manager
The network management software communicates with the agent
in the managed network element using special commands called
PDUs
The management information is carried over the network medium
using SNMP
Prepared by: Eng. Ali H. Elaywe
Continue
39
Figure 11 The relationship between the management
software, SNMP and the network element
Prepared by: Eng. Ali H. Elaywe
40
Topic 3: Management functions
 The network manager has information from many sources
including networked applications and elements on the
network. The functions undertaken by the manager were
discussed in Book M:



1- Accounting has, at the time of writing, yet to become a
major issue in LAN management. However, it is paramount in
WAN management, where the revenue comes from the use of
the network.
2- Security issues are beyond the scope of this text
The functions we want to concentrate on are those of 3configuration, 4- performance and 5- fault handling. These
may be grouped into two disciplines:


1- network design
2- network maintenance
Prepared by: Eng. Ali H. Elaywe
41
Sub-Topic 3.1: Network Design
 Many factors need to be considered in the design process if
optimal network performance is to be achieved. The following
lists the factors I think are important:





1- Simplicity: The network structure should be as simple as possible. It is
more likely to work, and will be easier to manage
2- Hierarchy: The design should be hierarchical, keeping layers for specific
purposes (as illustrated in Figure 12). Base layers should contain subgroups
that link via the layer above, and so on, within the hierarchy
3- Manageability: The network nodes should, where possible, support
SNMP. This will allow the network management system easy access to
the complete network
4- Future proofing: The design should allow for future expansion and not
just current needs
5- Robustness: The network must be stable under varying traffic loads,
with a minimum potential for failure. A contingency plan for recovery from
a catastrophic failure should be put in place
Prepared by: Eng. Ali H. Elaywe
Continue
42
Figure 12 Hierarchical model of an organizational network
(after Groom and Groom)
Prepared by: Eng. Ali H. Elaywe
Continue
43
 Issues of bandwidth requirements:




A choice has to be made regarding equipment types, data rates,
number and types of users within collision domains, etc
Table 1 gives typical amounts of data associated with common
activities. A standard Ethernet packet of 1500 bytes is assumed
Applications that access the WWW, involving multimedia, need a
much higher bandwidth than traditional text-based ones
The network manager must always be conscious of the fact that
demand will often outstrip the ability of the network to provide
sufficient bandwidth
Prepared by: Eng. Ali H. Elaywe
Continue
44
Activity
Byte count
Packet count
300 – 1500 bytes
1
200 bytes
1
File transfer
50 – 5000 Kbytes
30 – 3500
Print transfer
2.5 – 1000 Kbytes
2 – 700
60 bytes
1
E-mail message transfer
Request for an application
WWW page request
Table 1 Byte and packet counts for typical applications
(after Groom and Groom)
Prepared by: Eng. Ali H. Elaywe
45
Sub-Topic 3.2: Network Monitoring
 An Ethernet computer network is constructed from
several collision domains. The actual number of domains
will depend on several different factors
 Activity 7 (self-assessment)

Can you think of any factors that influence the number of
collision domains found within an Ethernet computer
network?

The factors that influence the number of collision domains found on
an Ethernet computer network include:
 1- Number of users
 2- Types of applications used on the network. Some applications will
involve greater data transfers than others
 3- The structure of the organization. High usage users may often be
situated in the same department
Prepared by: Eng. Ali H. Elaywe
46
Continue
 These domains are linked by network elements such as
hubs, bridges or routers. Monitoring the traffic at these
elements gives a measurement of the activity within and
between the collision domains
 Monitoring the traffic at various network elements
regarding the following statistics:


1- Normal operating conditions: measurements should include
data flows in each collision domain
2- Application usage: discovery of whether application usage is
between collision domains, leading to unnecessarily high traffic
levels within the network as a whole. An example might be high
usage of a remote server in one collision domain by users of a newly
installed application in another domain
Prepared by: Eng. Ali H. Elaywe
Continue
47

3- Inter-network traffic: in a network with many
remote sites, details of which routes over the
WAN are used for which applications, thus
defining normal usage
Prepared by: Eng. Ali H. Elaywe
48
Sub-Topic 3.3: Network Maintenance
 Network maintenance assures the best possible
service to the users
 There are two strands to network maintenance:


1- Preventative maintenance
2- Trouble management
Prepared by: Eng. Ali H. Elaywe
Continue
49
 1- Preventative maintenance:



is based upon the information derived from both network monitoring
and the results of routine tests, run to detect or correct problems
before faults are reported
These tests are usually performed at times when few users are on the
network, e.g. at night or weekends, so as not to interfere with day-to-day
work
The tests are designed to detect weaknesses in the network by
causing abnormal traffic levels. An example may be the discovery of a
bottleneck:
 This is a situation where high volumes of data in part of the network
can cause a degradation of the network performance. Once
discovered, steps can be taken to overcome the problem before
regular traffic levels rise to a level when the service degrades.
This avoids users being affected
Prepared by: Eng. Ali H. Elaywe
Continue
50
 2- Trouble Management



Trouble management associates an alarm, a visual and/or
audible indication that a problem exists, to a failure in
services and network resources
Alarms may be categorised as to their degree of
seriousness, from a warning of a possible problem (low
concern) to a complete failure of a collision domain (high
concern)
The operation of an alarm results in the initiation of
procedures to restore the network services to full operation

For example, an alarm can result from the inability of the network
to deliver packets if a domain fails. This may be due to a fault in
either network elements or media
Prepared by: Eng. Ali H. Elaywe
Continue
51

The Trap command, referred to in Sub-Topic 2.5,
causes a managed network element to communicate
a problem to the network management software,
which generates the appropriate alarm. In response to
the alarm, diagnostic tests can be used to identify
the problem and initiate procedures to replace faulty
components and restore the network to full service
Prepared by: Eng. Ali H. Elaywe
52
Topic 4: Manaqgement Tools
 SNMP enables network elements to present the
network manager with a vast amount of
information to identify usage trends and aid
longer-term growth planning. Use of this
information allows the network manager to be
proactive – (responding to potential issues
before they occur), rather than reactive –
(waiting for things to fail before taking action)
 The volume of information from network
management systems is huge
Prepared by: Eng. Ali H. Elaywe
53
Sub-Topic 4.1: Network Simulation
 Network simulators



are computer programs that ‘act’ like actual computer
networks, with outcomes that are comparable to
conducting experiments on a real network
They have recently been developed to import and use
SNMP data to aid the network manager with network
performance prediction
So Network simulators use analytical techniques, based
upon mathematical models, to represent the network
 Commercial network simulators are very powerful
programs. They are capable of simulating very
complex networks, as Figure 13 demonstrates
Prepared by: Eng. Ali H. Elaywe
Continue
54
Figure 13 A national network
Prepared by: Eng. Ali H. Elaywe
Continue
55
 Simulators, such as Opnet Modeler®, require the investment of
much individual effort plus supplier support in order to gain useful
results. However, the savings in time, money and the frustration
of the users make network simulators valuable tools. They are
capable of providing a high degree of network detail. Network
equipment manufacturers provide information about the operation
of their elements allowing them to be modelled by the simulator
 Furthermore, as Figure 14 illustrates, user tasks may be
specified for a particular simulation, along with categories of
users, as shown in Figure 15. This is a very powerful feature of
network simulation for it allows the network to be customised to
an individual organization including the pattern of usage by the
various users
Prepared by: Eng. Ali H. Elaywe
Continue
56
Figure 14 An example of simulated user applications
Prepared by: Eng. Ali H. Elaywe
Continue
57
Figure 15 An example of simulated users
Prepared by: Eng. Ali H. Elaywe
Continue
58
 You have the opportunity to use a network simulator,
OU LAN Simulator, which has been especially
developed for this module. Whilst it may not have the
sophistication of the commercial simulators, it does
allow you to experiment with a range of generic
network elements. This includes servers, printers,
hubs and bridges. Most importantly it includes the
ability to profile types of users and their associated
applications
Prepared by: Eng. Ali H. Elaywe
59
Sub-Topic 4.2: Media Management
 Once the network has been planned, decisions must
be taken as to the positioning of the various network
elements to ensure that the physical limitations of
the network media are not exceeded
 Activity 9 (self-assessment)

Which layer of the OSI Reference Model is concerned with
the physical medium of the network?

The bottom layer, i.e. layer 1, of the OSI reference model, is
the physical layer and is concerned with the physical medium
of the network
Prepared by: Eng. Ali H. Elaywe
Continue
60
 The physical layer covers all of the media: cables, connectors and
equipment, that link computers and other elements to the repeaters,
bridges and routers in the network
 All network media will have physical constraints on distance and
positioning. Care must be taken to ensure that the installation meets
the appropriate standards, the details of which are always changing as
new standards are introduced
 Whilst the network management system will probably display a
network map, it may not show the actual cable runs within the building
or site
 Cable-management software packages are available. They track all
the physical media including any unused cables that may exist. This
allows the exact position of suspect segments within the network to be
found
Prepared by: Eng. Ali H. Elaywe
61
Sub-Topic 4.3: Testing the network
 When installing a new service or after
troubleshooting procedures have identified a
possible media fault, devices such as cable testers,
similar to the one shown in Figure 16, can be used to
test the physical media


They are designed to test specific types of network media
and are usually handheld, portable devices
The tests are able to confirm that the link is functioning
correctly and not introducing any errors into the
transmitted data
Prepared by: Eng. Ali H. Elaywe
Continue
62
Figure 16 A handheld cable tester
Prepared by: Eng. Ali H. Elaywe
Continue
63
 Protocol analysers, sometimes referred to as
‘sniffers’, capture data transmitted between two or
more networked computers or elements
 Protocol analysers are able to decode the data
sufficiently to allow the contents of individual packets
to be displayed, as illustrated in Figure 17. This allows
you to view the packet data in a more easily readable
‘text-character’ format, shown on the right of Figure 17,
alongside the more usual hexadecimal display, shown
on the left
Prepared by: Eng. Ali H. Elaywe
Continue
64
Figure 17 The representation of a data packet by a protocol analyser
Prepared by: Eng. Ali H. Elaywe
Continue
65
 Protocol analysers provide information about the
traffic flow on the network from all elements including
those without management capability. They may be
programmed to recognise specific network addresses,
allowing the tracking of particular users or applications
 The protocol analyser usually consists of a PC (often
a laptop) running a protocol analyser application
(see Figure 18). The laptop allows the analyser
software to be physically carried to networked
computers and elements that do not support
management functions
Prepared by: Eng. Ali H. Elaywe
Continue
66
Figure 18 An example of a protocol analyser
Prepared by: Eng. Ali H. Elaywe
67
Topic 5: Preparation for future
 1) Must read Book E including the
practical work using the OU LAN
simulation and must read also Book W
 2) Do the activities on Book E
 3) Try to solve TMA03
 4) Prepare for the Midterm exam
Prepared by: Eng. Ali H. Elaywe
68