Basic Requirements for an Effective Network Management System
Managing a microwave radio network presents an interesting problem. The solution must deliver what network administrators and users want, and provide it in a manner that does not impact the performance of the radios. Network administrators want open systems solutions so that equipment from various vendors can interoperate together and be managed by a single and simple network management platform. Licensed microwave radios must also address the special nature of a wireless topology that is unlike a wireline LAN, as well as comply with FCC Part 101 rules in the United States.
This tutorial on network management will define what a network management system should accomplish and the guidelines that can be used to judge its effectiveness. It will also introduce the Simple Network Management Protocol (SNMP) which is the de facto standard for managing internet networks. SNMP in a Microwave Radio Network - Design requirements for an SNMP radio discusses a radio design that supports SNMP and the special nature of a wireless topology that requires a significant additional capability incorporated into the radio.
Requirements for Network Management
Most networks today, whether private networks or a part of big public networks such as the Internet, can be large and complex collections of expensive equipment. Often they are important to an organization's core infrastructure. This network of equipment must meet an end user's needs, and provide the "best" service at the "lowest" cost, and may generate significant revenue for the organization. In all cases, the network's continued operation and maximum utility is important.
Managing such a network can be a daunting task even when the objectives for managing such a system are known. These objectives might include getting better control over network assets, improving service to organizations that use the network, and reducing the downtime of any part of the network. Of course, controlling the costs of operating the network and its services is a given.
Many problems must be overcome when managing a network to accomplish these objectives. Networks tend to be large and complex, filled with equipment from many vendors; and there are many types of network managers, as well as proprietary solutions that are very popular and installed in many places. Therefore, as a network administrator, your "dream team" solution for the perfect network management would include the following features:
- easy to use
- manages heterogeneous devices and networks
- monitors the network's availability
- utility and performance
- proactively detects problems
- troubleshoots and isolates them quickly
- does all of this in a cost-effective manner
There are many vendor products that attempt to provide these network management solutions. They range from proprietary programs that can manage only one vendor's equipment to standards-based managers that can operate with many types of equipment from many vendors. However, no single product provides a "dream" solution, although the direction that most vendors and customers are headed is very clear. Customer requirements for simpler network management solutions are prompting equipment vendors to use open system standards that are easily understood and widely supported. They want equipment that is cost effective to implement and maintain, and are the best tools to manage heterogeneous devices and networks. The open system standards approach is proving to be the most flexible and cost effective way to implement a network management system.
Following is a discussion of the basic fundamental concepts of network management and the tools needed to meet customer requirements for managing their networks. Microwave radio networks have a unique history as managed networks, and this uniqueness is a driving force behind some significant developments that now make effective network management possible.
What is Network Management?
The network management is the collection of tasks performed to maximize availability, performance, security and control of a network and its resources. The International Organization for Standardization (ISO) Network Management Forum has divided network management into five functional areas:
- Fault Management
- Configuration Management
- Performance Management
- Accounting Management
- Security Management
Each of these functional areas has their own set of requirements, and an effective network management product should be able to perform in each area.
Fault management is the process of identifying and correcting network problems, otherwise known as faults. Faults typically manifest themselves as transmission errors or failures in the equipment or interface. Faults result in unexpected downtime, performance degradation and loss of data. Generally, fault conditions need to be resolved as quickly as possible.
Comprehensive fault management is the most important task in network management. Fault management tools can help increase the reliability of the network by quickly identifying the fault, and then help initiate the recovery process. The first step is to identify the fault, isolate the cause of the fault, and then, if possible, correct the fault. This three step process requires predetermining which faults should be managed and be given higher priorities than others, and then utilizing a set of tools to resolve the fault through the network manager. In the simplest case, an alarm is detected, and a maintenance technician is dispatched to locate and resolve the fault at the suspect location. Using more advanced tools, the network manager may be able to go many steps further to isolate and correct a fault from the manager location, and then return the network to normal operation without the user being aware of the failure.
The reporting of faults also falls under fault management. Universally recognized color such as red, green, and yellow can be used to indicate the status of a device - red for device in error, green for device with no error, and yellow for device with a minor error or possible error. Graphical presentations of a network map and front panels of the equipment are also used by the network manager to indicate status and operation of devices in the network.
Configuration management deals with the initialization, modification, and shutdown of a network. Networks are continually adjusted when devices are added, removed, reconfigured, or updated. These changes may be intentional, such as adding a new server to the network, or path related, such as a fiber cut between two devices resulting in a rerouted path. If a network is to be turned off, then a graceful shutdown in a prescribed sequence is performed as part of the configuration management process. The process of configuration management involves identifying the network components and their connections, collecting each device's configuration information, and defining the relationship between network components. In order to perform these tasks, the network manager needs topological information about the network, device configuration information, and control of the network component.
Performance management involves measuring the performance of a network and its resources in terms of utilization, throughput, error rates, and response times. With performance management information, a network manager can reduce or prevent network overcrowding and inaccessibility. This helps provide a more consistent level of service to users on the network, without overtaxing the capacity of devices and links. This form of management looks at the percentage of utilization of devices and error rates to help in improving and balancing the throughput of traffic in all parts of a network. Typically, some devices are more highly utilized than others. Performance monitoring give qualitative and time relevant information on the health and performance of devices so that underutilized devices are more fully utilized, and overtaxed devices are rebalanced. In a well-utilized network with healthy components, the error rates for packets traversing the network are down and response times are shortened.
Accounting management involves tracking each individual user's utilization of network resources for the purposes of allocation of resources and billing for their use of the network. This type of information helps a network manager allocate the right kind of resources to users, as well as plan for network growth. With the same information, the cost of transmitting messages across the network can be computed and billed to the user if the traffic was revenue bearing. This type of management involves monitoring the login and logoff records, and checking the network usage to determine a user's use of the network. In addition, access privileges and usage quotas can be established and checked against actual for accounting information.
Security management deals with ensuring overall security of the network, including protecting sensitive information through the control of access points to that information. Sensitive information is any data that an organization wants to secure, such as research documents, payroll data, and sales and inventory figures. Protecting sensitive data from unauthorized access is a common requirement. Security concerns can be assuaged with a well-designed and implemented security management system.
Security management controls access to the network devices and sensitive information through the use of devices such as passwords. This management also controls the form of sensitive data using methods such as encryption. There are many encryption techniques available for sensitive digital data such as public and private key encryption that have been in use for some time.
Each area of management in this five-part model is not exclusive of the others. It is typical for performance management to work in conjunction with fault and configuration management. For example, a monitored device interface that exhibits a slowly increasing error rate can be verified by an alarm query, and could then be reconfigured and bypassed before the fault affects traffic. Using these five management categories as guide lines, a more critical review can be made of any product's ability to manage or be managed as a network device.
Network Management Standards
Networks are growing at a rapid pace, and are usually built with products provided by many different vendors. Managing such a network is increasingly more difficult, involving multiple management tools and protocols to support different proprietary devices on the network. As a result, managers are demanding open systems solutions that use industry-accepted standards to reduce the number of different management systems. At the same time, vendors that comply to published standards can be assured that their products will interoperate together in a network.
Standards bodies have recently issued network management standards provide what vendors and network managers both need. The Internet Architecture Board (IAB), which is responsible for networking technology and protocols for the TCP/IP internetworking community, has created a standard network management protocol. These protocols are documented as RFCs or Request for Comments and have been made widely available.
The IAB recommends the Simple Network Management Protocol (SNMP) for use as a common network management protocol with TCP/IP-based networks. TCP/IP networks are the most popular types of network. SNMP, described by RFC 1157, became a full Internet Standard in May 1990, and is now in widespread use. Virtually every vendor of network equipment such as computers, workstations, bridges, routers, and hubs offer SNMP. Because of its widespread use as an open systems standard, new classes of equipment also support SNMP. That is a major reason for designing new microwave radios that support SNMP.
Network Management Architecture
Because networks that support the TCP/IP protocol suite are the most popular types of networks today, popular management systems for these networks are designed with the following key elements:
- Management station
- Management agent
- Management information base
- Network management protocol
- Any network management system for a TCP/IP network must have all four of these elements to function effectively.
A management station provides the interface for a human network manager to interact with the management system. This can be a workstation or a PC that executes a set of application programs for data analysis, fault management, data presentation, etc. The station is capable of translating the network manager's requirements into the actual tasks of monitoring and controlling individual network devices. In addition, it contains the database of information extracted from all the managed entities in the network.
There are many popular network management platforms on the market: Hewlett-Packard OpenViewTM, Cabletron SPECTRUM, SunNet Manager, IBM NetView, Harris Corporation FarScan, and Alcatel MCS-11. The platforms offered by HP, Cabletron and Sun are SNMP managers and can operate with any device that supports SNMP. HP OpenView is widely available and supported, and considered a de facto standard for management platforms. The platforms from Harris and Alcatel are proprietary solutions that do not comply to open industry standards and cannot interoperate with any network device other than their own products. This is a problem for network managers because the network may include equipment from different vendors that do not subscribe to the same proprietary platform. These devices cannot be managed by the management system; the manager is "blind" to these devices' activity and performance.
The management agent is the second active element in the management architecture. The agent is a software program in the network device that responds to requests for information or actions issued by the management station. The agent may also send the station unsolicited information, known as a Trap. Such a program is very much tied to the internal workings of the device. All devices in a network must have a management agent. Typically, an agent may be embedded or "native" to the device, or alternatively be a "proxy" agent for other protocols.
The third part of the architecture is the information that is exchanged between the manager and the agent; this is called the management information base or MIB. This information is a collection of objects or data values; each represents one aspect of the managed device. For example, the location of the device and the number of erred seconds in the last hour would be two different data values in the MIB. The structure and content of the MIB are standardized across systems of a particular class, such as a bridge MIB or DS-3 MIB. After a MIB is published as a standard, various vendors can build the same kind of equipment that complies with the MIB and be assured that they can be managed in a TCP/IP network.
The MIB structure is standardized in SNMP as a hierarchical tree. Additions to the tree can be easily accomplished, and traversing a tree to obtain specific information can be done very quickly. These are important features because they encourage the use of the MIB in the network management model and the creation of enterprise MIBs for vendors looking to support SNMP with their own products.
The last element of the model, the network management protocol, links the station and the agent by specifying the rules for communication. The protocol used for the management of TCP/IP networks is the Simple Network Management Protocol (SNMP). SNMP uses three simple commands to communicate: GET enables the station to retrieve data from the agent; SET enables the station to set a value at the agent, and TRAP that enables the agent to notify the station of an important event. Other protocols are available such as Internet Control Message Protocol (ICMP) and Simple Gateway Monitoring Protocol (SMGP) but they have limited functionality and only support a generic MIB. As a result, these two protocols are not widely used.
Introducing Simple Network Management Protocol (SNMP)
SNMP is the protocol for the management of TCP/IP networks. Two other specifications are necessary to form the foundation of the management system. Together, they are:
RFC 1157, A Simple Network Management Protocol: defines the protocol and architecture used to manage the MIB and its contents;
RFC 1155, Structure and Identification of Management Information for TCP/IP-based networks: describes how the MIB is defined;
RFC 1213, Management Information Base for Network Management of TCP/IP-based Internets: MIB-II: describes the contents of the MIB.
A set of related supporting RFCs are listed at the end of this section.
SNMP is an application layer protocol that is part of the Transmission Control Protocol/ Internet Protocol (TCP/IP) protocol suite. It is intended to operate over the User Datagram Protocol (UDP). The management station communicates management information using SNMP, which is implemented on top of the UDP and IP protocols, and a data link protocol such as Ethernet, Token Ring, or X.25. Likewise, the agent also must also implement SNMP, UDP, IP protocols.
SNMP has several key features that are important to understand. It is a transaction-based and datagram-oriented protocol. The protocol dictates that an action gets a response that ends the transaction. The response is a datagram or packet of bytes of prescribed size and format. Action by the management station, such as a GET command, will result in a datagram response from the agent. If the transmission fails to reach its destination address, the management station will retransmit again.
SNMP has a strong focus on Configuration and Fault management. It has fewer features for Security management; therefore, most SNMP management systems concentrate on alarm and status reporting, and use other methods for security. In addition, Traps sent by the agent to the management station are less reliable than GET responses because the UDP protocol is used. UDP does not guarantee delivery of the message. An alarm notification or trap from a faulty device may be lost because the datagram failed to reach the management station.
SNMP has many strong points in spite of the limitations mentioned above. Any vendor supporting SNMP must be aware of these limitations, and work around them to make their products competitive. SNMP is widely used by a number of vendors and its wide scale implementation makes it a de facto protocol for network management.
RFC 1155 Structure and Identification of Management Information for TCP/IP-based Internets. Rose, M.T.; McCloghrie, K. 1990 May.
RFC 1157 A Simple Network Management Protocol (SNMP). Case, J.D.; Fedor, M.; Schoffstall, M.L.; Davin, C. 1990 May.
RFC 1212 Concise MIB Definitions. Rose, M.T.; McCloghrie, K., eds. 1991 March. March 1991.
RFC 1213 Management Information Base for Network Management of TCP/IP-based Internets: MIB-II. McCloghrie, K.; Rose, M.T., eds.1991 March.