Electrical computers and digital processing systems: multicomput – Distributed data processing
Reexamination Certificate
1998-08-06
2002-07-02
Geckil, Mehmet B. (Department: 2152)
Electrical computers and digital processing systems: multicomput
Distributed data processing
C709S202000, C709S203000
Reexamination Certificate
active
06415314
ABSTRACT:
FIELD OF THE INVENTION
This invention relates to systems for network management, and more particularly to a system which allocates the management functions among different modules in a networking chassis.
BACKGROUND OF THE INVENTION
A computer network management system typically provides one or more of the following functions: monitoring activity on the network, detecting faults, generating alarms and/or isolating faults, allocating network resources, directing traffic, and determining or reconfiguring the network topology. As the complexity of computer networks increases, there is a growing need for improved management systems. In particular, there are concerns about a total or partial system “crash” (i.e., loss of network function) caused by a malfunction in the management system, the transmission and processing delays and reduction in memory space caused by the management operations themselves, and the inability to expand the network and/or major expense of replacing or upgrading the management system to accommodate a larger network.
In one prior art system, all management functions are provided on one module (“the management module”) which is plugged into the networking chassis. A “networking chassis” is a housing and backplane which receives “networking cards” that perform various networking services, such as repeating, bridging and routing. Each networking card or module includes its own microprocessor. In this prior art system, the “management module” has all of the hardware and firmware necessary to collect, store and process all of the data required to manage the system. This creates a serious problem if there is a malfunction in the management module and it needs to be pulled, i.e., there is nothing left to manage the system. To guard against this catastrophe, the user may purchase a spare module but this is an expensive method of insurance. Also, even during normal operation, consolidating all of the management functions in one module creates a potential bottleneck when there is an increasing level of transmissions and/or processing. Still further, the management module has a defined capacity and thus there is an upper limit on the amount of allowable network expansion (i.e., increase in the number of ports and/or traffic). For this reason, the purchaser of the system must decide whether to buy a larger management system than it presently needs but which will accommodate future expansion, or an adequate system which may have to be fully replaced if there is further expansion.
In another prior art system, each module in the chassis separately manages its own functions. In this case the chassis is merely a “housing” containing independent networking systems. In addition to the complexities of separate management, this system has problems similar to the “one management module” system in regard to the loss of network service accompanying each management malfunction, a potential bottleneck where each module must conduct its own management, and limited expansion capacity.
It is an object of the present invention to provide a new type of network management system wherein the system is managed “as a whole” but the management functions are “distributed” across the system.
It is an object to provide a plurality of modules in a networking chassis which together handle the management functions and wherein a malfunction in one module will not substantially effect the functions of the other modules and the overall management of the network.
Another object is to provide a system which permits ready expansion of the network without requiring replacement of the management system.
Another object is to provide a system which allows modification of the management functions without requiring replacement of the entire management system.
Still another object is to provide a system with a better allocation of resources for management functions in order to provide a system with greater throughput.
These and other objects of the present invention will be evident from the following summary and detailed description of select embodiments of the present invention.
SUMMARY OF THE INVENTION
A distributed chassis agent (“DCA”) for a network is provided which enables the chassis to be managed as a single system, and wherein any module can perform the management function or it can be performed by multiple modules simultaneously. The system scales to increasing module complexity and number as it spreads its workload across the modules contained within the chassis, discriminating against the most used modules. Using this system the degree of fault tolerance for the management of the chassis is equal to the number of modules contained within the chassis, as each module may be capable of performing the management function for the entire chassis.
The management function can be performed, for example, using the SNMP protocol which is part of the TCP/IP protocol suite. The management system is accessed via a network address which is known as the “chassis address.” The management function may be run on one or more modules within the chassis, but is always assessed via the same chassis address.
Three new functions of the chassis agent are: a) a discovery function conducted by each module to determine, store and send to the other modules information specific to that module, and to listen to the messages of other modules and store similar information regarding the other modules; b) an election function conducted by each module to determine which module(s) should conduct a specific management function; and c) distributed MIBs, wherein each object in the MIB has an identifier (known as an OID) which is registered both locally (i.e., on the module on which it resides) and remotely (on all other modules in the chassis) in a naming tree (MIB) located on every module, while the data for each object is stored only in one module. These and other new functions of the chassis agent are more fully described below.
One of the benefits of the new system is that it can operate without synchronization of the modules. This avoids the problems and complexities inherent in a synchronized system. Thus, each module can have its own clock and broadcast asynchronously (after a specified announcement period of, for example, one second), during discovery and other functions. Each module will continuously receive information from the other modules and update its own slot table of module information. The system is in a continuous state of “controlled instability” such that the necessary database updates and allocation of management functions are achieved within a few clock ticks by each of the modules.
In order for the networking chassis to function as a single system (i.e., in the view of the network and its users), the networking modules and other components (e.g., the power supply) within the chassis need to discover each other. Each module is required to keep track of the presence or absence of other modules and components within the same chassis, and of other operational parameters of each module/component. Module discovery is a continuous process, with each module issuing on a timely basis (order of seconds) an unsolicited message on the backplane of the chassis. The message contains basic information about the module, such as its slot ID within the chassis, internal management and external data link addresses, and the status of various objects on the module. Each module uses this information to build its own slot table containing the basic information about itself and similar information regarding the other modules. This information is used by a module to discover in which chassis it is currently installed. Once the module is discovered and entered into the slot table, the module may be polled for information about its resources. Each module includes its own processor (CPU), memory, and interfaces. The information in the slot table compiled by each module may include information concerning the type, speed and utilization of its CPU, the type, size and consumed amount of its memory, and the type and speed of its interfaces. The i
Arneson Dave
Dobbins Kurt
Fee Brendan
Mullaney Pat
Enterasys Networks Inc.
Geckil Mehmet B.
Wolf Greenfield & Sacks P.C.
LandOfFree
Distributed chassis agent for network management does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Distributed chassis agent for network management, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Distributed chassis agent for network management will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2911967