Method and apparatus for distributed object filtering

Abstract

Apparatus and method for distributed object filtering of objects transferred between a client and server. The objects may be alarms transferred between a network management server and one or more network management applications. The apparatus and method further includes applying a plurality of policy-based filters to the alarms. The filters may be named and stored in a database, and application of the policy-based filters may be scheduled for different times. The same policy-based filters may be applied to one or more multiple network management applications. The system and method provides increased performance and reliability of passing objects from a server to a client application.

Description

2. Field of the Invention

The present invention relates to distributed object filtering and more specifically to a method and apparatus for generating and receiving objects between a client and a server and applying filters to the objects.

3. Background of the Invention

Spectrum™ is a model-based network management system, sold by Cabletron Systems, Inc., Rochester, N.H., for maintaining and processing information pertaining to the condition of a communications network and providing the same to a user. For example, the Spectrum™ system will periodically poll a network device to request information, such as the number of packets sent on the network in a given time and the number of errors that occurred. If the error rate is above a predetermined limit, an error alarm is logged in a Spectrum™ system database, an object may be sent to the user interface to notify the network manager, and a message is sent to shut off the corresponding network device. These objects may be, for example, alarms that are passed from a server to a client. A server and client may be, for example, a network management system and an application program.

Alternatively, if no response was received from the network device when it was polled, the reason for the loss of contact should be determined so that appropriate action. Such as a service call, can be taken. In a network environment, loss of contact with a network device may be due to failure of that network device or to failure of another network device that is involved in the transmission of a message.

In many prior art network management systems, the network administrator was typically provided with a list of possible causes of a fault and was required to isolate the fault based on his experience and knowledge of the network. In Spectrum™, the system itself isolates network defaults using a technique known as Status Suppression. Spectrum™ maintains a database of models for each network device. When contact between a model and its corresponding network device is lost, the model sets a fault status and initiates the fault isolation technique. The model (first model) which lost contact with its corresponding network device (first network device) determines whether adjacent models have lost contact with their corresponding network devices; adjacent network devices are defined as those which are directly connected to a specified network device. If adjacent models cannot contact the corresponding network devices, then the first network device cannot be the cause of the fault, and its fault status in the first model will be overridden. By suppressing the fault status of the network devices which are determined not to be defective, the defective network device can be identified. Once the fault has been isolated, the condition of the defective device can be updated in the Spectrum™ database, a control message can be sent shutting off the defective device, and the network administrator can be notified via the user interface.

Spectrum™ associated SpectroGRAPH™ user interface provides a graphical view into the network models. An alarm log view, shown in FIG. 1 includes an area 120 for the listing of current alarms, and an area 122 for displaying information pertaining to a selected alarm. The user may click on a particular alarm in the listing of current alarms to obtain more information. A multi-function icon 124 representing the network device having a fault is displayed in area 122 , with one or more text fields 126 and 128 which provide information to the user regarding the cause of the alarm and the status of the device. By clicking on specified areas of the icon 124 , the user can obtain further information regarding the device for which an alarm is registered.

Another system for managing faults in large communications networks includes a what is referred to in the art as a “trouble-ticketing” system. This system provides a number of tools that can be used by network users, administrators, and repair and maintenance personnel. The basic data structure, a “trouble-ticket”, includes a number of fields in which a user can enter data describing the parameters of an observed network fault. A trouble-ticket filled out by a user may then be transmitted by, for example, an electronic mail system to maintenance and repair personnel. A trouble-ticket describing a current network fault that needs to be acted on is called “an outstanding trouble-ticket”. When the network fault has been corrected, the solution to the problem, typically called a “resolution” is entered into an appropriate data field in the trouble-tickets in memory and thus a library of such tickets is created, allowing users, administrators, and maintenance and repair personnel to refer to the stored completed trouble-tickets for assistance in determining solutions to future network faults. An example of a trouble-ticketing system is the ACTION REQUEST system, developed by Remedy Corporation, Mountain View, Calif., and sold by Cabletron Systems, Inc., Rochester, N.H.

ARS Gateway™ is a network management application sold by Cabletron Systems, Inc. which receives fault information from the Spectrum™ system and automatically generates a trouble-ticket that may be processed by the ACTION REQUEST system. This ARS Gateway™ may function as a client system that receives and processes objects. This system is further described in copending and commonly owned U.S. Ser. No. 08/023,972 filed Feb. 26, 1993 by Lundy Lewis, and entitled “Method and Apparatus for Resolving Faults in Communications Networks” and which is hereby incorporated by reference in its entirety.

Another application program for managing faults includes a business process management system. This system allows an administrator to manage one or more business processes, such as a payroll function. A business process may depend upon the network and various nodes in order to complete its goals. For example, when a business process such as payroll is described, the relevant system entities may include client workstations, computer servers, file servers, database servers, applications, peripherals, processes, and the network connecting these entities. Such a system is described in U.S. patent application Ser. No. 09/058,054, incorporated herein by reference.

The Spectrum™ system is described in U.S. Pat. No. 5,261,044 issued Nov. 9, 1993 to Roger Dev et al., which is hereby incorporated by reference in its entirety. The Spectrum™ network management system is commercially available and also described in various user manuals and literature available from Cabletron Systems, Inc., Rochester, N.H.

Other network management platforms and applications for the basic filtering of alarms which are commercially available include: (1) HP OpenView, 3000 Hanover Street, Palto, Calif. 94304; (2) Optivity, Bay Networks, 4401 Great American Pkwy., Santa Clara, Calif. 95054; (3) IBM Netview/6000, IBM Corp., Old Orchard Road, Armonk, N.Y. 10504; (4) SunNet Manager, SunConnect, 2550 Garcia Ave, Mountain View, Calif. 94043. and (5) Unicenter TNG. Computer Associates.

Some of the alarms passed between a server and a client system are routine or irrelevant, and the majority of alarms may be filtered out by the client application.

SUMMARY OF THE INVENTION

It is desired to increase the performance and reliability of passing objects from a server to a client application.

One embodiment of the present invention is directed to a method for distributed object filtering including: (a) generating alarms from multiple network management server; (b) assigning policy-based filters to both the network management servers and associated network management applications, and (c) applying the assigned policy-based filters to the alarms and for the alarms which pass the filters, generating an alarm notification and forwarding the same to the associated network management application.

In another aspect, an apparatus for distributed object filtering is disclosed comprising a database of policy-based filters; a user interface for assigning policy-based filters to both a plurality of network management servers and associated network management applications; a processor and a memory device containing a program of instructions for the processor which instructions include means for receiving alarms from the plurality of network management servers, means for applying, at both the plurality of network management servers and associated network management applications, policy-based filters to the alarms and generating an alarm notification for those alarms which pass the filters, and means for forwarding the alarm notification to the associated network management applications.

These and other features of the present invention will be more fully described in the following detailed description and figures.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is an example of an alarm log display provided by the Spectrum™ network management system.

FIG. 2 is a block diagram of an alarm notification manager for use with multiple network management servers and multiple network management applications.

FIG. 3 is a block diagram of a network management system within which an embodiment of the invention may be implemented.

FIG. 4 is a flow chart illustrating the application of policy-based filters to an alarm, and forwarding of the alarm which passes the filters to an application.

FIG. 5 is a block diagram of filtering of objects at the server and client.

FIG. 6 is a block diagram of an embodiment of filtering implemented on a client and server systems.

FIG. 7 is a block diagram illustrating filter processes.

FIG. 8 is a block diagram of work load balancing between client and server.

FIG. 9 is an example of an Association window display of the alarm notification manager.

FIG. 10 is an example of a new Association window display of the alarm notification manager.

FIG. 11 is an example of a Modified Association window display of the alarm notification manager.

FIG. 12 is an example of a Scheduler window display for the alarm notification manager.

FIG. 13 is an example of a Policies window display for the alarm notification manager.

FIG. 14 is an example of an Open Policy window display for the alarm notification manager. manager.

FIG. 15 is an example of an Add Filter Values window display for the alarm notification manager.

FIG. 16 is an example of an Alarm Age window display for the alarm notification manager.

FIG. 17 is an example of a New Policy window display for the alarm notification manager.

FIG. 18 is a block diagram illustrating two separate processes between the network management application and the alarm notification manager.

FIG. 19 is a block diagram illustrating a central processing unit and memory for use in this invention.

DETAILED DESCRIPTION

The present invention is directed to an alarm notification manager which receives alarms from multiple network management servers, allows an unlimited number of filters to be defined on either a client, a server or both, allows policies to be named and stored in a database, allows policies to be scheduled for different times, and allows the same policy to be applied to one or more network management applications.

As illustrated in FIG. 2 , a live network 10 is connected by links 11 to one or more network management servers 12 which monitor the network. The servers detect errors or faults on the network and send alarm information to the alarm notification manager 14 via links 13 . The alarm notification manager includes a policy database 16 , method for applying policies to alarms 18 , graphical interface 20 , and scheduler 22 . The manager applies policy-based filters to the alarm messages received from the servers, and for those alarms which pass the filter criteria, an alarm message is sent to the appropriate network management application 24 via links 23 .

In a specific embodiment described herein, a plurality of distributed SpectroServers™, part of the Spectrum™ system sold by Cabletron Systems, Inc., Rochester, N.H. are used to model the live network 10 , and several Spectrum™ applications received the filtered alarm messages from the manager 14 . These components have been implemented in the object-oriented programming language C++. However, the invention is not tied to any particular language nor to any particular products used in network management.

The Spectrum™ Network Management System

An understanding of the present invention is furthered by an understanding of the model-based network management system known as Spectrum™, which is described in U.S. Pat. No. 5,261,044, issued Nov. 9, 1993 to R. Dev et al., and hereby incorporated by reference in its entirety. The Spectrum™ network management system is commercially available and also described in various user manuals and literature available from Cabletron Systems, Inc., Rochester, N.H.

In summary, Spectrum™ is a system for maintaining and processing information pertaining to the condition of the computer network and providing the same to a user, the network including a plurality of network entities such as computer devices and software applications being executed on such devices. The system includes a virtual network machine, comprising a programmed digital computer, wherein a program is implemented using an object-oriented programming language such as Java, C++, Eiffel, SmallTalk, or Ada. The virtual network consists of interrelated intelligent models of network entities and relations between network entities, including means for acquiring network data pertaining to the condition of a network entity from the corresponding network entity. The virtual network further includes means for maintaining objects which include network data relating to the corresponding network entity and one or more inference handlers for processing the network data, the inference handlers being responsive to changes occurring in the same and/or a different object. The network data can then be transferred to a user interface coupled to the virtual network machine, for supplying the network data to a user.

Thus, the models are implemented as software “objects” containing both “data” (attributes) relating to the corresponding network entity and one or more “inference handlers” (functions) for processing the data. See Grady Booch, “Object-Oriented Analysis and Design With Applications,” 2nd Edition, Benjamin/Cummings Publishing Co., Redwood City, Calif., Chapter 2, 1994. The inference handlers are initiated by predetermined virtual network events, such as a change in specified network data in the same model, a change in specified network data in a different model, and predefined events or changes in models or model relations. Information pertaining to the condition of the network entity can be obtained from the network entity by polling the same, can be automatically received from the network entity (without polling), or can be inferred from data contained in other models. An alarm condition may be generated when the network data meets a predetermined criteria Events, alarms and statistical information from the virtual network are stored in a database and are selectively displayed for the user.

The data in the Spectrum™ database may be used for generating topological displays of the network, showing hierarchial relationships between network devices. isolating a network fault, and reviewing statistical information.

Spectrum™ allows for collective management of autonomous local area networks (LANs), with equipment from different vendors. It complies with the current Simple Network Management Protocol (SNMP) standards, and can also accommodate other standard and proprietary protocols. The virtual network machine preprocesses the raw information coming from the network devices in order to construct a model of the network's current status and performance characteristics. Network elements that cannot be directly communicated with (e.g., cables and buildings) can infer their status from the status of the devices connected to (or contained within) them. The virtual network machine provides a consistent interface for management applications to access any of the information in the model and thereby provides these applications with a unified view of the network.

Spectrum's™ associated SpectroGRAPH™ user interface provides a highly graphical multi-perspective view into the network model. SpectroGRAPH™ enables the user to navigate through a landscape in which cables, networks, local area networks and rooms show up as icons, and which icons indicate the health and performance characteristics of those elements. These icons can be further queried for additional information. SpectroGRAPH™'s main function is to visually present to the user the model within the virtual network machine. It allows the user to navigate freely within the network model, only limited by the access rights assigned by the network administrator. The information can be accessed at varying degrees of detail, from a macro overview, to the devices and cables which connect them. In addition to its navigation functions, SpectroGRAPH™ provides an alarm management facility, an event log window, a reporting facility, a find facility, and other features.

The above description of the Spectrum™ system provides a context for an understanding of the present invention.

The Alarm Notification Manager

The following definitions are helpful to an understanding of the present invention:

SANM        SPECTRUM ™ Alarm Notification Manager
Policy      A set of criteria which a given alarm must satisfy in order
            to be passed to the application with which the policy is
            associated. A policy may consist of one or more filters.
Filter      A set of filter parameters and associated filter values.
            Multiple filters define multiple sets of values for the filter
            parameters.
Filter      A data type such s model name or IP subnet for which the
Parameter   user can specify a value or list of values. SANM provides
            the user with a fixed list of filter parameters.
Association When the user associates a policy with an application, he
            is specifying the filter criteria that SANM should apply to
            the alarms it sends to the application.

A filter may include a list of filter parameters and a list of associated filter values. A user (of a network management application) specifies the value(s) that each filter parameter can take in order for a given alarm to pass the filter criteria. The following is a list of representative filter parameters:

model name

model type name

device IP subnet

device location

alarm severity

alarm age

SpectroSERVER host name

landscape name

alarm cause

The value for each of the above filter parameters would be received from Spectrum™, except for the alarm age parameter. The alarm age parameter is used internally by SANM and specifies the length of time that it should hold an alarm before sending it to an application. If the alarm is cleared by the Spectrum™ system during this time, it is not sent to the application. This feature may be used to filter out transient alarms.

Each filter value also has a corresponding flag which indicates whether it should be negated. For example, if the negate flag is set for a model type name value of Hub_CSI_IRM3, this filter value states that all alarms for models NOT of type Hub_CSI_IRM3 should pass.

More complex filtering can be achieved by defining multiple filters within a policy. Each filter specifies a separate set of filter values.

SANM performs a logical AND of all the filter criteria within a filter and performs a logical OR between all filters within a policy.

For example, a policy contains two filters as follows:

Filter 1
Model Type: Rtr_Cisco
Landscape:  wiz
Filter 2
model Type: Rtr_Wellfleet
Landscape:  brat

SANM would apply this policy to a given alarm as follows:

IF the alarm has:
model type Rtr_Cisco AND landscape wiz
OR
model type Rtr_Wellfleet AND landscape brat
THEN send the alarm to the application

Each filter also contains a filter tag, which is a text string that the user enters. This tag, which is included in the alarm notification, identifies which filter(s) passed and can be used by an application to perform routing of alarms.

For example, a different user name can be entered in the filter tag field of each filter, so that if the criteria in one filter pass, the application will notify a particular user, whereas if the criteria in another filter pass, the application will notify a different user. If multiple filters pass, a list of corresponding filter tags is sent in the alarm notification.

Another example of the SANM filtering mechanism is shown in FIG. 4 . In this figure, the criteria listed within each filter are the criteria for which values have been specified by the user. It can be seen from this example that all filters are applied in parallel to a given alarm (i.e., a logical OR is performed between filters). However, all criteria within a given filter must be satisfied for the alarm to pass the filter (i.e., a logical AND is performed between the criteria within a given filter). Since, in this example, the alarm passes the criteria in filters 1 and 3 , an alarm notification containing filter tags “A” and “C” is sent to the application.

Policies and the associations between policies and applications are stored in the SPECTRUM™, database. This means that the same policies are available to any client machine running SANM. It also means that the policy names contained in event messages logged by SANM have significance to all client machines using the same SPECTRUM™ database.

One embodiment of the invention aims to increase the performance and reliability of passing objects from a network server to a client application, particularly in cases in which a large number of objects are considered for passing but not all need to be passed. The objects that are passed from a server to a client are generally alarms (or events). Many of these alarms are routine or irrelevant, and the majority of alarms are eventually filtered out and discarded by the client application. The task of filtering alarms may be distributed over both the server and client. Filtering is allowed to take place on the server side and/or the client, depending on available resources, in order to maximize performance and reliability of client/server applications.

Existing methods of event passing allow filtering to take place on the client side. FIG. 3 shows a configuration in which the SANM client receives alarms from several SpectroSERVERS. Note that filtering is not performed on the server side. It is the responsibility of SANM to filter the alarms it receives and pass the allowable alarms to the BP representation for further processing.

FIG. 3 shows a high-level drawing of the invention as applied to a network with three domains. Network 110 is a large network, e.g., a global network. Domain 112 is one domain of the network, e.g. the LA domain. Spectrum 140 monitors Domain 112 via communication link 130 and produces intra-domain alarms, i.e., those alarms that are specific to Domain 120 . SAMN 160 collects intra-domain alarms from each Spectrum via communication link 150 and forwards them to a BP representation 180 via communication link 170 . BP representation 180 correlates the set of intra-domain alarms from each Spectrum and produces higher-level, BP alarms. These alarms may then be forwarded back to Spectrums via the Spectrum Command Line Interface 200 . i.e., higher-level BP alarms produced by BP representation 180 are forwarded to CLI 200 via communication link 190 , and CLI 200 forwards the higher-level BP alarms to appropriate Spectrums via communication link 210 .

The disadvantages of filtering only on the client are two-fold:

(i) An extra burden is placed on the client. If the client is busy performing other tasks, and is also flooded with events or alarms upon which to perform the filtering task, then the performance and reliability of the client may be reduced.

(ii) The volume of traffic on the medium between the client and server may be overstressed, and thus also reduce overall performance and reliability of the client/server application.

Performance and reliability of a client/server application is increased by distributing the filtering requirement over both the server and client, depending on resources available and preferences. The invention thus allows more flexibility in tuning such applications for optimal performance. The following examples show percentages of filtering distributed over a client and a server, and show circumstances in which the distribution would be appropriate.

Server	Client	Appropriate when:
100	0	no bandwidth to spare, no client resources to spare
0	100	bandwidth to spare, but no server resources to spare
25	75	preference for lessening stress on server
75	25	preference for lessening stress on client

FIG. 5 shows an exploded view of a client and single server connection, where components 140 , 150 , and 200 are similar to the like-numbered items in FIG. 3 . Filtering may be performed (optionally) at both the client and the server. Note that if all filtering is placed on the client, the client may become a bottleneck. Likewise, if all filtering is placed on the server, the server may become a bottleneck.

U.S. patent application Ser. No. 08/412,955, incorporated herein by reference, discloses a system having user interfaces and mechanisms that are used to set the filtering configuration on the client side. In this system, the user interfaces and mechanisms may be used to control the SANM client only. According to one embodiment of the present invention, similar interfaces and mechanisms may be used at both the client and server to control distributed filtering for both the client and server.

The user interface may be used also to control the parameters upon which filtering is based. For example, in the SANM alarm filtering application shown in FIG. 6 , the parameters may include model 301 , model type 302 , alarm condition 310 , and alarm age 311 . A even distribution (50—50) of filtering over these parameters is the placement of model and model type filters on the server side, and alarm condition and alarm age on the client side.

The invention may implement from a principle of robot reasoning called the multi-loop architecture (MLA) for robot monitoring and control. The right-hand side of FIG. 7 shows the basic structure of the MLA. Each level of the MLA is a separate control loop that corresponds to a specific class of problems, where problems are partitioned ad assigned to levels according to the amount of time and type of information required to solve them.

For example, the short-term abstraction/correlation/decomposition loop at the lowest level provides quick reaction to problems, by-passing upper level control mechanisms. In the network management domain, such tasks might include intelligent routing and temporary disconnection to a busy host. These tasks are event-based. The medium-term loop provides reaction to more complex problems and operates on increasingly abstract input such as alarms. Tasks on this level might include intra-domain alarm correlation in a single domain, where some alarms are real and others are apparent and the task is to distinguish the two and suppress all apparent alarms. The top level provides reaction to problems that require more time. A good example of a task of this kind is BP alarm correlation notification.

Alternatively a task may be described as a purely abstraction task without any form of correlation or decomposition of information. The invention described herein is of this type. FIG. 7 shows the relation of concepts described in this invention to intelligent monitoring and control (see the enclosed components within the dotted circle).

The distribution of work load between the Client and the Server (or alteratively the work load between two or more peer intelligent agents assigned to cooperatively perform a task) may be decided a-priori by a human agent and can be set through a User Interface. One embodiment of the invention includes the methodology that will allow intelligent agents (residing at the Server and the Client ends) to decide the work load distribution dynamically through information exchange between the involved agents. FIGS. 5 and 8 describe the methodology considering two agents (running at the Server and the Client machines).

The server side agent 320 runs on the server machine 140 . Similarly the Client side agent 330 runs on the client machine 200 . The server and the client machines are connected together by the transmission medium 150 . In one embodiment, the server agent and the client agent may communicate to each other via a distributed blackboard 380 . The blackboard 380 may be, for example, a storage area on the distributed network. Both the agents post their part of the information regarding their resource availability, their state of business on the blackboard. A separate agent or either of the server agent or the client agent can keep track of the load characteristics (bandwidth availability, packet drop rate) of the transmission medium.

Both the server agent and the client agent operate on a set of agreed-upon load distribution policies 370 . These policies are definable and changeable by a human agent through a policy definition/alteration user interface 375 .

The load distribution policy is based on a “composite metric of suitability” of any agent to perform a task at a given moment. The metric is based on other component metrics such as:

speed of the machine on which the agent is running (ultra Sparc vs. 200 MHz Pentium machine vs. Some Symmetric Multi-processor machine),

memory and disk space available on the machines

current cpu availability

current number of “threads” running

other considerations

The “composite metric of suitability” for each of the agents to perform the current task (such as filtering an object based on N characteristics) is calculated by assigning some suitable weights to each of these component metrics. Both the server and the client agents may calculate their “composite metric of suitability” at a regular interval of time and post it to the black-board. When the time to perform a task arrives, the most recent “composite metric of suitability” numbers of the server agent and the client agent are looked up from the black-board and the work load is distributed between the server agent and the client agent is distributed according to the load distribution policy. For example, us assume the “composite metric of suitability” for the server agent is 42 and that for the client agent is 37 and the load distribution policy requires the load to be distributed in proportion to the values of the “composite metric of suitability” for each of the agents and the task at hand is to filter an object based on 7 characteristics. The load distribution in this case may be performed as follows,

Number of characteristics filtered by the server=42*7/(42+37)=3.7=4

Number of characteristics filtered by the server=37*7/(42+37)=3.2=3

It is evident from the above method that the load distribution can change dynamically if either of the server or the client machine characteristics change or their state of busyness changes.

1. Alarm Notification

After an application has registered with SANM to receive alarms, an alarm notification is sent to that application each time an alarm is received from SPECTRUM™ that passes the criteria specified in the policy associated with that application. The information contained in each alarm notification consists of the real-time values of each filter parameters, plus the values of the following parameters:

model handle

model type handle

model condition value

model security string

alarm ID

alarm time

alarm probable cause

alarm status

event message associated with alarm

assigned repair person

user-clearable flag

One exception to this is that an IP subnet address may be specified as a filter criterion, but the full IP address of the device that created the alarm is passed in the alarm notification.

A notification that an alarm has been cleared or updated is sent to an application when SANM receives such a notification from SPECTRUM™, but only if the alarm which is being cleared or updated was initially sent to the application when it occurred (i.e. it passed the filter criteria for that application).

2. Configuration Tool

The SANM Configuration Tool enables the user to define Alarm Notification Policies and to associate these policies with the applications that use SANM.

The Configuration Tool is invoked by selecting Alarm Notification Manager from the asterisk menu of SpectroGAPH™.

2.1 Associations Window

When the Configuration Tool is invoked, the first window to appear is the Associations window, shown in FIG. 9 . This window displays a list of the currently defined SANM applications and the policy that is associated with each of them.

A new association is created by selecting New from the File menu. This brings up the New association window shown in FIG. 10 .

An existing association is modified by selecting the association and then selecting Modify from the File menu. This brings up the Modify Association window shown in FIG. 11 .

An existing association is deleted by selecting the association and then selecting Delete from the File menu. The selected association is deleted after the user confirms the operation in a Confirmation Dialog window (not shown).

The modification of an existing association can be scheduled by selecting the association and then selecting Schedule from the File menu. This brings up the Scheduler window shown in FIG. 12 .

All currently defined policies can be viewed by selecting Policies from the Tools menu. This brings up the Policies window shown in FIG. 13 .

2.2 New Association Window

The New Association Window is illustrated in FIG. 10 . In this window, a policy is selected from the list of available policies and the application name is entered. When OK is pressed, the window disappears and the new association appears in the Associations window (FIG. 9 ).

2.3 Modify Association Window

The Modify Association window is illustrated in FIG. 11 . In this window, the user picks a policy from the list of available policies to associate with the selected application (SpectroPHONE™ in this example, available from Cabletron Systems, Inc.). Pressing OK makes this window disappear and the modified association is displayed in the Associations window (FIG. 9 ).

2.4 Scheduler Window

The Scheduler window is illustrated in FIG. 12 . Pressing the Associate button brings up the Modify Association window illustrated in FIG. 11 . In the Modify Association window, the user picks a policy from the list of available policies to associate with the selected application (SpectroPHONE™ in this example). In the Scheduler window, the user then presses the Frequency button to specify the frequency of the association. The Frequency options are: Once, Hourly, Daily, Weekly and Monthly. The information in the area below the Frequency button changes depending on what frequency option is selected as follows:

The Once option allows the user to specify the month, day and start-time.

The Hourly option allows the user t specify the number of minutes after each hour.

The Daily option allows the user to specify the time.

The Weekly option allows the user to specify the day of the week and the time.

The Monthly option allows the user to specify the day of the month and the time.

Once the desired scheduling options have been selected, pressing the Add button inserts the scheduling information into the Scheduled Entries portion of the window. Further entries can be added by repeating the previous steps. Entries can be modified and removed by selecting them and using the Modify and Remove buttons.

2.5 Policies Window

The Policies Window is illustrated in FIG. 13 . This window shows all currently defined policies.

A new policy is created by selecting New from the File menu. This causes the New Policy window ( FIG. 17 ) to appear.

An existing policy is viewed and modified by selecting the policy and then selecting Open from the File menu. This causes the Open Policy window ( FIG. 14 ) to appear.

An existing policy is deleted by selecting the policy and then selecting Delete from the File menu. The selected policy is deleted after the user confirms the operation in a Confirmation Dialog window (not shown).

2.6 Open Policy Window

The Open Policy window is illustrated in FIG. 14 . This window shows all the filters that make up the policy. In the example shown in FIG. 14 , Filters 1 and 2 are visible but subsequent filters can be viewed using the scroll bar on the right of the window. Similarly, the other filter parameters for Filter 1 and their associated values can be viewed using the scroll bar below the Filter 1 filter parameters.

To modify the displayed policy, Edit must be selected from the File menu. The View item in the menu bar then becomes Edit. Once in Edit mode, multiple values for a particular filter parameter can be deleted or negated by selecting the values and pressing the Delete or Negate button. Values can be added for a particular filter parameter by pressing the filter parameter button (e.g., Landscape or Model Type). This brings up a separate window containing a list of available values from which multiple values can be selected. An example of this window is shown in FIG. 15 .

Filter parameters may be added to a filter by pressing the Parameter button within the filter. A pop-up menu appears containing all eight filter parameters. However, those filter parameters which are already present in the filter are greyed-out and cannot be selected. Selecting one of the available filter parameters from this menu causes the new filter parameter and associated value box to appear in the filter.

The alarm age for a particular filter can be modified by pressing the Age button in the Open Policy window. This brings up the Alarm Age window shown in FIG. 16 . The values for the Hours and Minutes fields initially contain the values from the Age text field in the open Policy window. These values can be modified using the up and down arrow buttons for hours and minutes.

A filter tag can be modified in the Open Policy window by typing directly into the Tag text field of a filter.

A new filter may be added to the policy displayed in the Open Policy window by pressing the Create Filter button. This will cause a new filter with no filter parameters to be added to the end of the list of filters.

An existing filter may also be duplicated. To do this the filter to be duplicated must first be selected by clicking within the filter label field (e.g., the area around the label Filter 2 ) and then pressing the Duplicate Filter button. Doing this causes a new filter, containing the same filter parameters and values as the selected filter, to be added to the end of the filter list. This new filter can then be modified.

After modifying a policy, Save can be selected from the File menu to save the modified policy under its existing name, or Save As can be selected to save the modified policy under a different name.

The information in the Open Policy window can be printed by selecting Print from the File menu.

2.7 New Policy Window

The New policy Window is illustrated in FIG. 17 . The operations that can be performed in the New policy window are the same as those performed in the Open Policy window (FIG. 14 ). No filter parameters initially appear within Filter 1 , therefore the first operation that needs to be performed is to select a filter parameter by pressing the Parameter button within Filter 1 . All filter parameters are available from the pop-up menu at this point because the filter does not yet contain any filter parameters.

A new policy is saved by selecting Save As from the File menu and entering the name for the policy in a dialog box.

3. Integration of SANM and Application

A developer would use the following interface to integrate an application written in C or C++ with the Spectrum™ alarm mechanism.

An application using SANM to receive alarm notifications and to clear/acknowledge alarms requires two separate processes, as illustrated in FIG. 18 .

As an example of how these two separate processes would be used in an application, the ARS Gateway™ product would use Process 1 to receive filtered alarms from SANM. format them into Trouble Tickets and put them into the ARS Database. Process 2 would be used when a user viewing a particular Trouble Ticket pressed a clear or acknowledge button in the Trouble Ticket.

Two different programming paradigms are required for the two application processes that use SANM:

For the process that receives alarm notifications from SANM, an asynchronous callback paradigm is used. This means that when the application code registers with SANM to receive alarms, it hands program control over to SANM. When SANM needs to send an alarm notification to the application, the application receives a callback from SANM. This process is terminated by sending it a TERM (terminate, 15 ) signal.

For the process that clears or acknowledges alarms, however, a synchronous paradigm is used. This means that the application code in the is process has program control. When this application code makes a call t the SANM API to clear or acknowledge an alarm, the call blocks the application until it is finished.

3.1 Definitions and Data Structures

All definitions and data structures are contained in the SNAM header file sanm.h and are described below.

The prototype for the application's callback functions is defined as follows:

typedef void (*SANMb) (struct SANM_Alarm_Notify *);

All the data in an alarm notification is contained in the SANM_Alarm_Notify structure, which is defined as follows:

struct SANM_Alarm_Notify{
char          *Model_name;
SANMUlong     model_handle;
char          *model_type_name;
SANMUlong     model_type_handle;
int           condition_value;
char          *security_string;
SANMUlong     alarm_ID;
SANMTimestamp alarm_time;
SANMUlong     cause_code;
char          *probable_cause;
char          *alarm_status;
char          *event_message;
char          *repair_person;
char          *IP_address;
char          *location;
SANMUlong     severity
SANMUlong     alarm_age;
char          *SpectroSERVER_host;
char          *landscape;
SANMBoolean   user_clearable;
char          *filter_tag;

All errors and warnings are defined in the enumeration SANM_error as follows:

enum SANM_error
{
SANM_RETURN_OK,
SANM_INVALID_ALARM,
SANM_INVALID_LANDSCAPE,
SANM_ALARM_NOT_CLEARABLE,
SANM_REGISTER_ERROR
}

3.2 Functions

The functions that make up the SANM C/C++ API are described in the following sections in manual page format.

3.2.1 SANMInit
NAME
SANMInit - initialize interaction with SANM SYNOPSIS
#include “sanm.h”
SANM_error SANMInit ( char *application_name, SANMBoolean
rcv_or_clr);
DESCRIPTION
SANMInit serves to initialize the program for interaction with
SANM. This function should be called from within both application processes before any
other function in the SANM API.
INPUT ARGUMENTS
application_name        the name which must be used by the user
                        to identify this application when using
                        the Configuration Tool to associate a
                        policy with it.
rev_or_clr              a flag which indicates whether this
                        process is going to receive alarm
                        notifications or clear/acknowledge
                        alarms. The flag can take either of the
                        following two values:
                        SANM_RCV_ALARMS
                        SANM_CLR_ALARMS
RETURN VALUES
status                  The return value will be one of the
                        following values:
                        SANM_RETURN_OK
3.2.2 SANMRegister
NAME
SANMRegister - register with SANM
SYNOPSIS
#include “sanm.h”
SANM_error SANMRegister (SANMCb set_cb
                        SANMCb clear_cb,
                        SANMCb update_cb);
DESCRIPTION
SANMRegister registers the application to receive alarm notifications
from SANM. By calling this function, the application hands program
control over to SANM until one of the application's callback functions
is called.
INPUT ARGUMENTS
set-cb                  the name of the function that SANM will
                        call in order to send an alarm notification
                        for a new alarm. All applications must
                        pass a valid function for this parameter.
clear_cb                the name of the function that SANM will
                        call in order to send an alarm notification
                        for a cleared alarm. This parameter can
                        be NULL if the application does not want
                        to receive notifications for cleared
                        alarms.
update_cb               the name of the function that SANM will
                        call in order to send an alarm notification
                        for an updated alarm. This parameter can
                        be NULL if the application does not want
                        to receive notifications for updated
                        alarms.
RETURN VALUES
status                  In normal operation, this function will
                        never return. However, if it fails, one of
                        the following errors will be returned:
                        SANM_REGISTBR_ERROR
3.2.3 SANMClear
NAME
SANMClear - clear an alarm
SYNOPSIS
#include “sanm.h”
SANM_error SANMClear ( SANMUlong alarm_ID char *landscape );
DESCRIPTION
SANMClear clears an alarm in SPECTRUM. An application
can only clear alarms for which it received notifications from
SANM. Also, the user-clearable flag must have been set to
CLEARABLE in the alarm notification
INPUT ARGUMENTS
alarm_ID                the ID of the alarm to be cleared
landscape               the landscape that generated the alarm
RETURN VALUES
status                  The return value will be one of the
                        following values:
                        SANM_RETURN_OK
                        SANM_INVALID_ALARM
                        SANM_INVALID_LANDSCAPE
                        SANM_ALARM_NOT_CLEARABLE
3.2.4. SANMack
NAME
SANMAck - acknowledge an alarm
SYNOPSIS
#include “sanm.h”
SANM_error SANMAck ( SANMUlong alarm_ID, char *
landscape);
DESCRIPTION
SANMAck acknowledges an alarm in SPECTRUM. An
application can only acknowledge alarms for which it received
notifications from SANM.
INPUT ARGUMENTS
alarm_ID                the ID of the alarm to be
                        acknowledged
landscape               the landscape that generated the
                        alarm
RETURN VALUES
status                  The return value will be one of
                        the following values:
                        SANM_RETURN_OK
                        SANM_INVALID_ALARM
SANM_INVALID_LANDSCAPE

The present embodiments may be implemented in a general purpose computer 70 as shown in FIG. 19 . The general purpose computer may include a computer processing unit (CPU) 71 , memory 72 , a processing bus 73 by which the CPU can access the memory, and interface 74 to the rest of the alarm notification manager.

In alternative embodiments, the invention may be a computer apparatus which performs the functions of any of the previous embodiments. Alternatively, the invention may be a memory, such as a floppy disk, compact disk, or hard drive, that contains the computer program or data structure, for providing to a general purpose computer instructions and data for carrying out the functions of the previous embodiment.

Having thus described certain particular embodiments of the invention, various modifications will readily occur to those skilled in the art which are intended to be within the scope of this invention. Accordingly, the foregoing description is by way of example only, and not intended to be limiting.

Claims

1. A method of distributed object filtering comprising the steps of:(a) generating alarms from multiple network management servers; (b) assigning policy-based filters to both the network management servers and associated network management applications; and (c) applying the assigned policy-based filters to the alarms and for the alarms which pass the filters, generating an alarm notification and forwarding the same to the associated network management application.

2. The method according to claim 1, wherein the step of assigning includes assigning the policy-based filters according to a load distribution policy.

3. The method according to claim 2, wherein the load distribution policy is based on at least one of a plurality of metrics including processor availability, memory availability, disk space availability, and processing capability.

4. The method according to claim 2, further including a step of determining the load distribution policy based on a busyness of at least one of the network management servers and a busyness of at least one of the network management applications.

5. The method according to claim 2, further including a step of determining, for one of the network management servers, a first metric that represents a suitability of the one network management server to perform a filtering operation.

6. The method according to claim 5, further including a step of determining, for one of the network management applications, a second metric that represents a suitability of the one network management application to perform the filtering operation.

7. The method according to claim 2, further including a step of periodically scheduling policy-based filters.

8. The method according to claim 4, further including a step of dynamically determining the load distribution policy.

9. The method according to claim 6, further including a step of communicating the first and second metrics between the one network management server and the one network management application.

10. The method according to claim 6, further including a step of distributing filtering operations in proportion to the first and second metrics.

11. An apparatus for distributed object filtering comprising:a database of policy-based filters; a user interface for assigning policy-based filters to both a plurality of network management servers and associated network management applications; a processor and a memory device containing a program of instructions for the processor which instructions include: means for receiving alarms from the plurality of network management servers; means for applying, at both the plurality of network management servers and associated network management applications, policy-based filters to the alarms and generating an alarm notification for those alarms which pass the filters; and means for forwarding the alarm notification to the associated network management applications.

12. The apparatus according to claim 11, wherein the processor and memory device containing a program of instructions further include:means for assigning the policy-based filters according to a load distribution policy.

13. The apparatus according to claim 12, wherein the processor and memory device containing a program of instructions further include:means for determining for one of the network management servers, a first metric that represents a suitability of the one network management server to perform a filtering operation.

14. The apparatus according to claim 13, wherein the processor and memory device containing a program of instructions further include:means for determining, for one of the network management applications, a second metric that represents a suitability of the one network management application to perform the filtering operation.

15. The apparatus according to claim 14, further comprising a storage area, wherein the one network management application and the one network management server exchange the first and second metrics.

16. The apparatus according to claim 13, wherein the processor and memory device containing a program of instructions further include:means for distributing filtering operations in proportion to the first and second metrics.