Management Architecture and Standards

IACT 418 IACT 918Corporate Network Planning

Gene AwyzioSpring 2001

Objectives

Recognise the protocols used in management of a telecommunications network with reference to the OSI stack, in particular SNMP ASN 1 MIB

Identify databases and their use to manage the telecommunications network

Objectives

Understand what is involved in change control, configuration management and configuring a networkDemonstrate appropriate practical and problem solving skills

The OSI Stack

Proposed by the International Organisation for Standardisation (ISO) and other standards bodiesIts a layered approach to network protocols involving encapsulation of packets at each level of the stack and the sub-layers

The OSI Stack

Each layer has its own functionalityAll layers are likely present at the end user’s host machine or Front End processor (FEP) or customers equipment, but …

The OSI Stack

Inside the network you may have only the first two or three layers A router mostly has up to the network layer A bridge uses the data link layer (sublayers

LLC [Logical Link Control] and MAC [Medium Access Control])

The network management protocols operate at the APPLICATION layer, like TELNET and FTP

The OSI Stack

Transport

Data Link

Physical

Network

Session

Presentation

Application7

6

5

4

3

2

1

The OSI StackLayer

Name Encapsulation

Component

7 Application Data Software

6 Presentation

Data Software

5 Session Data Software

4 Transport Segments Router

3 Network Packets, Datagrams

Router

2 Data Link Frames NIC. ( LLC & MAC )

1 Physical Bits NIC.( Physical ) Hub, Repeater. DCE & DTE

Network Management Software Architecture

A network management system consists of incremental hardware and software additions (scalable) implemented among existing network elementsThe software used to accomplish the network management tasks resides in the host computers and communications processors: Front End Processors (FEP) terminal cluster controllers Bridges Routers

Network Management Software Architecture

A network management system allows viewing of the entire network as a unified architecture: With addresses and labels assigned to each

network element With specific attributes of each network

element and its connectivity known to the system

The active elements of the network provide regular feedback of status information to the network control centre

Network Management Software Architecture

A generic network management structure might be as given in the following figure:

Network Management Software Architecture

NMA

NME Appl

NMA

Comm

NME Appl

NMA

Comm

NME Appl

NMA

Comm

OS

NME

Comm

Network Management Software Architecture

Each node contains a collection of software devoted to an appropriate network management task Referred to as the network

management entity ( NME)

Network Management Software Architecture

At least one (and probably more ) host in the network is designated the network control host or managerAs well as the software for a NME, these hosts include extra software called the network management application (NMA)

Distributed Network Management

Centralised network management systems are devolving into distributed network management systemsDue to the same factors are involved that have lead to the distributed networks Low cost of hardware / PC’s High powered PC’s Proliferation of LAN’s

Distributed Network Management

A distributed management system replaces the single network control centre with interoperable workstations located on LANs distributed throughout the organisationThis gives departmental (local) level managers the tools they need to maintain responsive networks, systems and applications for their local end users

Distributed Network Management

But… how do we avoid anarchy?

Network

Distributed Network Management

NetworkManagement server

MIBManagementApplication

Management server

MIBManagementApplication

Element Manager

NetworkNetwork

Element Manager

Management clients (PCs, workstations)

Network Resources(servers, routers, hosts) with management agents

Distributed Network Management

The distributed system has the following benefits: Network management traffic is minimised Distributed management offers greater

scalability adding capability only involves adding another

workstation in the appropriate location The use of multiple networked stations

eliminates the single point of failure represented by having only one central manager

Proxies

Proxies may be required when: Older network elements exist in the network

that do not use the protocols implemented by the management servers

Small systems that would be unduly burdened (too many management packets would flow on the system) to support a full blown NME (agent)

Modems and multiplexers which do not support additional software / agents

SNMP

The term Simple Network Management Protocol (SNMP) actually refers to a collection of specifications that include: the protocol itself the definition of data structures associated concepts (eg proxies)

We will now look at a brief overview of the SNMP setup, the most prolific method of network management today

The History

SNMP’s development follows a similar pattern to the development of TCP/IPARPANET evolved during 1970’s, into a set of protocols approved by the internet architecture board (IAB) {www.iab.Org}

The History

Standards were issued as RFC’s (Request for comment) and were adopted by the military in the United States in their procurement policies, forming the core of the TCP/IP protocol suite, which was adopted by vendors for non military activities, and grew dramatically during the 1980’s

The History

Many still predict that OSI standards will replace TCP/IP, but this has not occurred at this time. Why? Possible reasons are: TCP/IP suite is mature, and adapts to

interoperability issues via the IAB / IETF TCP/IP provides a high level of functionality -

from the majority of users points of view The international standards have been slowly

developed (OSI), while providing a richer functionality they will also require more effort in implementation and conformance testing

The History

As TCP/IP developed little thought was given to network management, especially in the early days where many programmers and protocol designers would modify the network as requiredThrough the late 1970’s there were still no network management protocols, but the one tool used for management that was available was the Internet Control Message Protocol (ICMP)

Internet Control Message Protocol:

ICMP Provides a means for transferring control

messages from routers and other hosts to a host to provide feedback about problems in the networking environment

Most useful message is the echo / echo reply message pair - provides a mechanism to test if communication is possible between two network elements / entities on the TCP/IP network

Also provides a mechanism for measuring delay across the connection using time stamp and time stamp reply

Internet Control Message Protocol:

The ICMP messages can be used with IP header options such as source routing and record route ( a trace route feature) to develop simple but powerful management toolsThe most useful example of this is the widely used PING (Packet INternet Groper)

Internet Control Message Protocol:

PING can perform a variety of functions Determine if a particular network element

can be addressed Verifying that a network can be addressed Observe variations in round trip times Observe datagram loss rates, which can

help isolate points of congestion and high error rates / points of failure

Internet Control Message Protocol:

This was satisfactory up until the late 1980’s when exponential growth in the internet required more powerful network management capabilityIn November 1987, the Simple Gateway Monitoring Protocol (SGMP) was issued, providing a way to monitor gateways on the internet

Internet Control Message Protocol:

At this time three approaches were proposed or emerged: High Level Entity Management System

(HEMS) Simple Network Management Protocol

(SNMP) CMIP (Common Management

Information Protocol from OSI) over TCP /IP (CMOT

Internet Control Message Protocol:

In 1988 the IAB reviewed these proposals and approved SNMP for a short term solution and CMOT as the long range solution Both SNMP and CMOT were to use the same

database of managed objects

There would be only a single Structure of Management Information (SMI - the basic

format conventions for objects) Management Information Base (MIB - the actual

structure or schema of the database)

Internet Control Message Protocol:

In 1989, after it was found impractical to have the same set of managed objects, the IAB relaxed its condition of common SMI / MIB and allowed the two protocols to develop independentlyFreedom from OSI constraints led to rapid SNMP development

SNMP Related Standards

Three foundation specifications of SNMP are: Structure and Identification of Management

Information for TCP/IP-based networks (RFC 1155)

Management Information Base for Network Management of TCP/IP-based Internets: MIB-II (RFC 1213)

Simple Network Management Protocol (RFC 1157)

The SNMP Network Management Architecture

The model used for network management of TCP/IP includes the key elements: Management station Management agent Management Information Base Network management protocol

 Management Station

Typically a stand-alone network element, but may be a capability on a shared system (eg mainframe)Serves as the network interface for the human network manager into the network management system

 Management Station

 As a minimum the network management station will have: A set of management applications An interface The capability of translating the network

manager's requirements into the actual monitoring and control of remote elements in the network

A database of information extracted from the MIBs of all the managed entities in the network

 SNMP standards only cover points 3 and 4

Management agent

The management agent Responds to requests for information

and actions from the management station

May asynchronously provide the management station with important but unsolicited information

Managed objects

 Resources in the network may be managed by representing these resources as objects Each object is a data variable that represents one

aspect of the managed agent

The collection of objects is referred to as a Management Information Base (MIB) The MIB functions as a collection of access points

at the agent for the management station

Management Information Base

The objects are standardized across systems of a particular class For example, a common set of objects

is used for the management of various bridges

 A management station performs the monitoring function by retrieving the value of MIB objects

Management Information Base

A management station can Cause an action to take place at an

agent change the configuration settings at an

agent by modifying the value of specific variables

The management station and agents are linked by a network management protocol

SNMP - the Management Protocol Used for TCP/IP

 SNMP includes the following key capabilities: Get Set Trap

The standards do not specify The number of management stations The ratio of management stations to agents

SNMP - the Management Protocol Used for TCP/IP

In general, it is prudent to have at least two systems capable of performing the management station functionsAs SNMP is simple it can handle many agentsSNMP is designed to be an application-level protocol that is part of the TCP/IP protocol suite which operates over the user datagram protocol (UDP)

SNMP - the Management Protocol Used for TCP/IP

SNMP - the Management Protocol Used for TCP/IP

SNMP - the Management Protocol Used for TCP/IP

From a management station, three types of SNMP messages are issued on behalf of a management application: GetRequest GetNextRequest SetRequest

SNMP - the Management Protocol Used for TCP/IP

The first two are two variations of the get functionAll three messages are acknowledged by the agent in the form of a GetResponse message, which is passed up to the management application

SNMP - the Management Protocol Used for TCP/IP

An agent may issue a trap message in response to an event that affects the MIB and the underlying managed resources - this is received by the managerSNMP relies on UDP, which is connectionless so SNMP is itself connectionless ie each exchange is a separate transaction between a management station and an agent

Trap - Directed Polling

Preferred strategy is: A management station can poll all of the

agents it knows for some key information Once the baseline is established, the

management station refrains from polling Each agent is responsible for notifying the

management station of any unusual event

Trap - Directed Polling

These events are communicated in SNMP messages known as trapsOnce a management station is alerted to an exception condition, it chooses to take the appropriate action

Trap - Directed Polling

Trap-directed polling can result in substantial savings of Network capacity Agent processing time

Reduces unnecessary polling of agents by managers thus reducing management induced network traffic

Limitations of SNMP

SNMP may not be suitable for the management of very large networks Each agent needs to be polled and

generates trap traffic SNMP is not suited to retrieving large

volumes of data such as a entire routing table

SNMP traps are unacknowledged meaning the agent generating the trap does not know if the manager received it

Limitations of SNMP

Basic SNMP standard only provides trivial authenticationSNMP does not directly support imperative commands with parameters, conditions, status and results

Limitations of SNMP

SNMP MIB model is limited not supporting sophisticated management queries based on object values or typesSNMP does not support manager to manager communications ie no mechanism for one manager to find out about another network managers, managed network elements