Management of a Network

User Generated

fvfu1990

Computer Science

Description

TASK - 1

a) Summarize the attached research article in One to two pages. the summary should include the main idea presented in the paper with appropriate explanation. b) Write your reflection on the network management architecture from the research paper.(advantages, disadvantages, etc) (Proper Referencing and documentation is required).

TASK - 2

As a network manager for a small/medium company, you are required to be aware of the health of your LAN and help in taking decisions regarding its growth, security and utilization rules by remotely monitoring some probes. You started your Remote monitoring project by setting a number of objectives:

1. The current utilization percentage of the bandwidth and its trend.

2. Create history reports

3. Know which conversations (communicating hosts) are generating most traffic.

4. Define the settings to capture a specific type of frames.

5. Set a number of alarms raised by some critical states.

6. Monitor the traffic for some upper layer’s traffic.

Task 2.a:

For each objective described in this task, you are required to find the RMON group and tables that will be pulled to get or set from it some parameters as well as all objects that will be used. For some objectives, you need to set first the control table associated.

Example: To determine the traffic statistics with one host, we need to access the tables: “hostControlTable” and “hostTable” in the Host group.

The objects that can be invocated: In “hostControlTable”: hostControlTableSize and hostControlLastDeletedTime, hostControlStatus,hostControlIndex. In “hostTable” : hostAddress, hostCreationOrder and hostIndex.

Task 2.b:

you need to give an example for the selected objects with a clear explanation.

Example: In “hostControlTable”: hostControlTableSize=10: The number of hostEntries in the hostTable is 10. It is calculated by

counting distinct MAC addresses in the frames read.

hostControlLastDeletedTime=0;is the value of sysUpTime in the System group of mib-2 at which a row in the hostTable was deleted. If no deletion was done than this number is set to 0.

For all other objects described in task a), you need to provide an example and its explanation.

***how many groups in the Object

***if one, write about it and write about the tables in it?

***if more than one group, write about them and write about the tables included in each group

Task 2.c:

You are required to conduct a comparison research about 5 Open source tools for Network monitoring and write a report based on it. The report must include the features and limitations of each tool.

*** Citation and Referencing Using CU Harvard Style is Required.

*** Support your answers with diagrams or graphs.

*** Word Count: Min 3000 words

User generated content is uploaded by users for the purposes of learning and should be used following Studypool's honor code & terms of service.

Explanation & Answer

Hi! am almost through, one more thing could you send me the e-book there's something I need to confirm....
Your assignment is complete, if you have any queries shoot me a message and hang tight, I'll assist you in a couple minutes or asap. :-)AM HERE FOR YOUHave a great day ahead.

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/3195162

The OSI network management model
Article in IEEE Communications Magazine · June 1993
DOI: 10.1109/35.212418 · Source: IEEE Xplore

CITATIONS

READS

105

529

1 author:
Yechiam Yemini
Columbia University
121 PUBLICATIONS 4,197 CITATIONS
SEE PROFILE

All content following this page was uploaded by Yechiam Yemini on 19 March 2014.
The user has requested enhancement of the downloaded file.

The OSI Model

The OSI Network Management
Model
Balancing the responsibilities of OSI’s agents and platforms — and
their interaction protocols — is complex, but OSI helps by offering
functions lacking in Internet’s SNMP.
Yechiam Yemini

T

YECHIAM YEMINI is a
member of the Computer
Science Department at
Columbia University, where
he presently serves as the
director of the New York
State Center of Advanced
Technology in computer and
information systems.

20

here were dramatic shifts in the
structure and role of networked systems within enterprises in the past
decade. From isolated data-processing islands, networked computing systems have grown into complex
mission-critical enterprise-wide systems. The network, the computer, and the very enterprise
rapidly are becoming indistinguishable. These
changes lead to significant risks and cost exposure associated with operations. For example,
failures of a bank network can paralyze its operations, delays of security trades through brokerage
system bottlenecks can cost in dollars and customers,
and loss of hospital lab reports can prevent timely
diagnosis and care. The goal of network management technologies is to reduce the risks and costs
exposure associated with operations of enterprise
systems.
Management systems are responsible to monitor, interpret, and control the network operations. A typical management system is depicted in
Fig. 1. Vendors equip their devices with agent
software to monitor and collect operational data
(e.g., error statistics) into local databases, and/or
detect exceptional events (e.g., error rates exceed
threshold). Management platform workstations
query device data, or obtain event notifications
through management protocols. The management platform supports tools to display the data
graphically, interpret it, and control operations.
This management paradigm is platform centered.
Management applications are centralized in platforms, separated from the managed data and
control functions in the devices. Platform-centered management reflects older network environments where devices lacked resources to run
management software, management data and
functions were relatively simple, and network
organizations could devote the personnel needed to
handle operations. The implications of these assumptions and their validity for current networks will
be considered later.

0163-6804/93/$03.00 1993© IEEE

The main challenge of management standardization is to develop conventions to support integrated management of heterogeneous networks.
Platform-centered management requires a few standards. First, access by platforms to multivendor
devices must be unified through a standard management protocol. Second, the structure of the agent’s
management databases, manipulated by the protocol, must be standardized. Together, these
standards permit a platform to access and manipulate managed information at multivendor device
agents. The OSI and Internet management models seek to standardize both areas.
Merely moving management information from
devices to platforms, however, is insufficient to eliminate the curse of heterogeneity. Two additional barriers to integrated management arise: platform
and semantic heterogeneity.
Platform heterogeneity means that management
applications must be replicated for each major
platform. For example, a device vendor wishing
to offer six applications over five platforms may need
to develop and maintain 30 product versions. Therefore, a number of recent consortia (e.g., OSF,
XOPEN, POSIX) are pursuing management
platform standards.
Semantic heterogeneity arises when different
devices use different information to represent
similar network behaviors. A management application program requires a uniform semantic
model of the managed information it processes.
It is necessary to standardize the very meaning of
managed information. Various IEEE and CCITT
protocol committees pursue this challenge, building
managed information standards for protocol entities.

Why Is Management Difficult?
onsider an example of a network “storm” to illustrate management complexities. Storms
C
involving rapid escalation of cascading failures
are not uncommon in networks. Figure 2 depicts

IEEE Communications Magazine • May 1993

Management Platform

Management Protocol

Agent

Agent

Agent

Agent

Enterprise Network Computing System
■ Figure 1. Architecture of a network management system.
a T1 link multiplexing a large number of connections
(e.g., X.25 virtual circuits, or TCP) to a server/host.
Suppose a long burst of noise disrupts the link
causing packet loss (Fig. 2a). Logical link level
protocols (above the physical layer) invoke automatic retransmission. They result in a burst of retransmission tasks at the interface processor queue
(Fig. 2b) loading its queue and leading to its
thrashing. Higher layer transport entities timeout and respond with a burst of corrective activities (e.g., reset connections). This burst processing
of communications at host CPUs (Fig. 2c) leads
to their thrashing, too.
Generally, protocol stack mechanisms handle
lower-layer problems through corrective actions
at higher layers. These mechanisms can escalate
the very problems they intend to solve.
How can such complex network fault behaviors be monitored, detected, and handled? Suppose that relevant operational variables (e.g., T1
bit-error rates and the size of the interface processor queue) can be observed as depicted in
Fig. 3. The storm formation could be detected
from the correlation of the sudden growth in
error rates and the resulting growth in queue size.
What management information should be
used to capture these behaviors? The Simple
Network Management Protocol (SNMP) uses a simple model for the structure of managed information (SMI) [2] involving six application-defined data
types and three generic ones.
Temporal behaviors are described in terms of
counters and gauges. An error counter represents
the cumulative errors (integral) since device
booting (the area under the error rate curve in
Fig. 3). A gauge can model the queue length. The
values of these managed variables can be recorded in an agent’s management information base (MIB)
[3], where they can be polled by a platform. An
error counter, however, is not useful for detecting
rapid changes in error rates to identify a storm. A
platform must sample the counter frequently to estimate its second derivative, leading to unrealistic
polling rates.
OSI management uses an object-oriented model
of managed information [9, 10]. The behaviors of
interest: noise, errors, and queue length are different forms of a time series. A generic managed

IEEE Communications Magazine • May 1993

Resets ( c )

Retransmissions ( b )
Noise
Server

T1
Interface

Frame loss ( a )

■ Figure 2. Formation of a network storm.

Queue length
Error rate

Error
count

Time

■ Figure 3. Temporal behaviors correlation.
object (MO) class may be defined to describe a
general time series. This MO may include data
attributes of the time series and operations
(methods, actions) to compute functions of the time
series (e.g., derivatives). This MO also may provide generic events notifications (e.g., whenever some
function of time series exceeds threshold). The generic time series MO class may be specialized to
define MO subclasses to model the bit-error rate
of the T1 link and the queue length of the interface processor. A management platform can create instances of these MOs, within device agents’
databases. The device agent can monitor the respective network behaviors and record the respective

21

O SI
management
communications require
connectionoriented
transport
and rely on
the OSI
application
layer
environment.

Ma...


Anonymous
Excellent resource! Really helped me get the gist of things.

Studypool
4.7
Trustpilot
4.5
Sitejabber
4.4

Similar Content

Related Tags