Electrical computers and digital processing systems: multicomput – Computer network managing – Computer network monitoring
Reexamination Certificate
1999-03-22
2002-08-06
Harrell, Robert B. (Department: 2152)
Electrical computers and digital processing systems: multicomput
Computer network managing
Computer network monitoring
Reexamination Certificate
active
06430617
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to network management systems and in particular relates to methods and associated systems that dynamically measure the ability of a computer system to support data collection for analysis by network management system applications.
2. Discussion of Related Art
It is common for computer systems to be linked via communication media for exchange of information and to distribute computational tasks among a plurality of such interconnected computer systems. Such collections of interconnected computer systems and/or the communication paths that interconnect them are often referred to as networks. Computing tasks may be distributed over the network of computer systems by application of distributed computing techniques known in the art. Data and computing resources such as storage devices, printers, and other devices may be shared over such networks so that a user at any of the computing systems of the network perceives that he/she is locally attached to the shared resources and data.
Use of networks has grown rapidly as computing applications have evolved. The Internet is an example of a network that has grown to worldwide proportions with millions of users and their computer systems interconnected at any given moment. Other networks are contained within a single organization—an enterprise. For example, a single corporation may use a network to connect all elements of their business. Such an enterprise-wide network would connect all systems in a building and may extend to other buildings and other remote sites even around the world. Though the Internet is not centrally managed by a single entity, enterprise computing environments are typically managed by a single central entity—a network manager—or at most a relatively small group of network managers cooperating to manage the computing resources of the enterprise through the network.
Small networks (i.e., a few computing systems connected within the same office area) are easily managed by a network manager by physically visiting each computing system to monitor performance, diagnose problems, configure systems, etc. However, large networks (i.e., a large number of computing systems physically dispersed with respect to one another) present logistical problems for network managers. It is difficult or impossible to manage such networks where physical presence of a network manager is required at each computing system of the enterprise.
Network managers in such large computing enterprises often use special purpose computer programs to monitor operation and performance of the enterprise network. Such programs are often referred to as network management systems (or NMS).
An NMS is a combination of hardware and software used to administer various aspects of network operation by controlling the configuration of equipment used in the network's infrastructure. Most NMSs are operable on a computing system (node) of the network to gather data regarding performance of the network and/or to diagnose operation of the network. Such NMSs are generally a collection of one or more computer application programs that exchange messages with other computing systems and communication devices (nodes) on the enterprise network to achieve the desired data acquisition and diagnostics. In all but the most trivial networks an NMS also performs critical data consolidation and filtering functions that allow the network manager to focus on the information of the greatest importance to him: data that reveals existing or pending problems with the network's performance.
Because an NMS is critical to the operation of most networks many NMSs consist of one or more software packages that execute on dedicated host computer systems, or workstations. In smaller networks an NMS may consist of software executing on a workstation that is also used for other business application purposes. Due to the heavier load of managing larger networks, NMSs on larger networks tend to run on one or more workstations dedicated to the NMS operations.
NMSs usually include component applications that execute periodically or even continuously to monitor various aspects of network performance over time. Such NMS applications might do things like periodically check to ensure that all portions of the network are operable or collect network data that reveals which nodes utilize the network most heavily. If the data collected by these NMS applications is to accurately reflect the state of the network over time, each NMS application must be able to execute its functions completely and without restriction.
Yet such NMS applications compete with one another, other network management software, and even completely unrelated software programs for the limited resources of the host computer system on which they execute.
Every workstation's physical memory and processing capacity are limited and therefore valuable resources. Further, the workstation's operating system restricts other resources needed to collect network management data such as virtual memory (secondary storage capacity used to extend the effective size of main memory), file descriptors (data structures/objects used by application programs to manipulate files stored on persistent storage of the computing system), and network input/output (the system's capacity to handle network traffic).
Generally speaking, assuming a given NMS package, the amount of each resource needed to effectively monitor a network increases roughly in proportion with the size of the network. The inevitable outcome faced by a network manager with a growing network is that the network outstrips the monitoring capabilities of their NMS as operable on a particular computing system. This problem is increasingly common as networks evolve from simple topologies (i.e., a single shared communication path such as a single Ethernet segment) to more complex ones such as micro-segmented (desktop-switched) topologies.
In such a situation the network manager must invest more money to maintain an NMS that provides adequate control of the network. One option is to purchase a new NMS package that can better handle large topologies; this can require the network manager to conduct a time-consuming survey and evaluation of available products, expend several thousand dollars, and endure a steep learning curve to realize the benefits of the new software. Another option is for the network manager to keep their current network management software and simply upgrade the workstation on which the software executes; this may be less expensive than buying new software but is generally a stopgap solution. Eventually, the enterprise network will grow to a point where the NMS capability will require more resources than presently available.
Some present NMS applications utilize configuration parameters and models of the systems on which they run to limit their demands for resources. Some NMS applications, for example, inspect certain statically configured aspects of their host system and adjust their behavior accordingly. These inspections usually involve some combination of system parameters such as the type and configuration of network media and protocols, the type and speed of network interface adapters, the central processor type and speed, and the system memory. Often these parameters are inspected upon installation or initial configuration of the NMS application or perhaps inspected once when the NMS application is first started. However, it is impractical and imprecise to inspect all of the applicable settings on a host at a given time in order to project the system resources that will be available to an individual network management application during its subsequent execution. Even if all of the statically configured aspects of the system that affect NMS application performance were examined and their cumulative effect accounted for, it is impossible to predict the impact that other applications executing on the same host will have on the network management application's interaction with
Britt Steve Van
Klinger Linda Ann
Harrell Robert B.
Hewlett-Packard Co.
LandOfFree
Methods and systems for dynamic measurement of a... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Methods and systems for dynamic measurement of a..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Methods and systems for dynamic measurement of a... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2883979