Cisco Network Mgmt Protocol FAQ: The Basic Ingredients of Network Management
Q1. Name the two contexts in which the term agent is used in network management.
Answer: 1) The agent role—that is, the role of the managed network element, as opposed to the managing system.
2) The management agent—that is, the software component of a network element that implements a management interface and handles management communication for the network element.
Q2. Compare the manager/agent and client/server paradigms. What are the commonalities, and what are the differences?
Answer: In both cases, communication between the two roles is asymmetric. Manager and client roles are analogous, in that both own the initiative in the relationship and both send requests to the system in the other role. Agent and server roles are analogous, in that they are both subservient to the requests sent by the system in the other role. However, the manager/ agent paradigm differs from the client/server paradigm because general managers manage many agents (each one managed by only one or very few managers), whereas servers serve many clients.
Figure: Manager/Agent Reference Diagram
Q3. The chapter stated that a network element can contain more than one management agent and that a management agent can contain embedded management intelligence. Taking these statements literally can lead to the conclusion that the same management intelligence might have to be implemented redundantly in a network element, once for each management agent. Clearly, this would be a wasteful approach. What would be an appropriate refinement of the model of a management agent?
Answer: The management intelligence should be shared between different management agents, making the same intelligence accessible across several management interfaces. One way would be to move the intelligence into the layer of the real resource. However, embedded management intelligence is not part of the function of the network and should, hence, not be considered a real resource. This implies that the implementation of management agents itself results in a layered subsystem architecture inside the network element. This architecture needs to separate functions and “intelligence” that are generically applicable across management interfaces (and that should therefore be implemented only once) from functions that are genuinely tied to the particular management interface and the way the agent represents management information.
Q4. Explain the term MIB—what does the acronym stand for, what is it, and who provides it?
Answer: MIB stands for Management Information Base. It is a conceptual data store that represents an abstraction of a network element for management purposes. It is provided by a system in an agent role.
Q5. Name one difference between a MIB and a database.
Answer: The MIB represents a view of real resources, not a set of passive data items that need to be maintained from the outside.
Q6. Tell whether the following statement is true: “If a network is required to have availability of 99.999 percent, its management systems need to also be 99.999 percent available.” Why or why not? Please elaborate. For extra points, factor in the influence of the type of application that the management system is used for.
Answer: A network functions independently of its need to be managed. Accordingly, the 99.999 percent availability should not be affected if the management system’s availability is less. Therefore, to achieve five nines, five nines availability of the management system is generally not required. However, we need to distinguish between the types of management applications that the management system supports: Availability of a management system used for monitoring the network should not be allowed to drop much below five nines because the lack of monitoring capability could adversely affect the network’s availability. On the other hand, availability of a management system used to add new users to a network is less critical. Finally, management systems used to collect accounting data that is used to bill for services by a service provider does need to be as highly available as the network—not for reasons of network availability, but because any loss of accounting information can result in loss of revenues because customers cannot be billed.
Q7. Management traffic is different from other communication traffic, in that the NE itself is a destination and originator of traffic. However, it is not the only type of traffic for which this is true. Name an example of other network traffic that the NE does not just switch or route, but actively participates in.
Answer: Control and signaling traffic, such as traffic from routing protocols.
Q8. What could be the most important reason for using a dedicated management network instead of a shared one?
Answer: The most important reason generally concerns reliability. In case of network failures and congestion, it is important to communicate with network elements to diagnose what happened and apply reconfigurations to remedy the situation remotely. Without a dedicated management network, in those cases it will be hard to get through to the network element, just when it is needed the most. A second reason concerns avoiding interferences with other network traffic, with which management traffic would compete for networking resources.
9. Which other term do service providers use to refer to management systems?
Answer: Operational support systems (OSS).
10. Would you expect a management system to provision services to be located at a NOC or at a Central Office? Why?
Answer: At a NOC. The NOC is the Network Operations Center from which network management involving operations personnel takes place. Central Offices, on the other hand, resemble remote branch offices in an enterprise. They terminate local lines and might not even be staffed.