Method and apparatus for processing requests for statistics in a communication network
A method and apparatus of initiating a plurality of requests for statistics to be sent to an associated plurality of nodes in a segment of a communication network is provided. One or more requests are associated with an individual node. Each node has one or more node attributes. The method includes defining a sequence for initiating the requests to be sent to the nodes. The sequence is based on a value of a selected node attribute of each node. The method also includes initiating the requests according to the sequence.
This application is a continuation of U.S. patent application Ser. No. 09/833,531, filed Apr. 11, 2001, and incorporated herein by reference.
FIELD OF THE INVENTIONThe invention relates generally to data communications, and more particularly to a method and apparatus for processing requests for statistics in a communication network.
BACKGROUND OF THE INVENTIONCommunication networks comprise a plurality of nodes, or switches, which are interconnected to form a web of nodes. Users communicate with other users in the network by transmitting messages and data through the network. Data is routed or switched through a path of connected nodes from the node associated with the sender to node associated with the recipient.
Each node may maintain a set of statistics relating to the network traffic passing therethrough. Statistics include performance, usage or billing data collected from a communication network. The statistics may be used for billing purposes, evaluation of network performance at a node and fault diagnostics that may be present about a node. Typically, statistics are tracked using software modules which track aspects of the statistics in counters or registers internal to the nodes. The counters may be embodied in hardware or software using techniques known in the art. Generally, to conserve resources, the counters maintain only recent data. Further, the counters may be configured to be reset after a predetermined time period or interval. Typical reset intervals are 5 minutes, 15 minutes, 60 minutes and 24 hours. The counter occupies a portion of memory on the node to store data from the previous interval while compiling data for the current interval. At the end of the current interval, interval data is written over the memory contents for the previous interval. The counter for the current interval is reset and then begins to track data for the current interval. As such, statistics must be collected in every interval in order to retain timely statistics.
In prior art systems, there are statistics collection systems for a communication network which typically send statistics requests to the network using the known simple network management protocol (SNMP) or other protocols appropriate for the nodes. The statistics collection system typically communicates with the network using a separate device, a statistics collection unit. The statistics collection unit is typically external to the network and is connected to the network using a dedicated network management functionality communication link.
It will be appreciated that statistics collection requires resources. Processing a request for statistics collection from the statistics collection unit by a node requires that the node utilize its processing resources, thereby diverting resources from processing switching or routing traffic. Accordingly, nodes typically are not optimized for statistics collection processing and generally commit only limited resources thereto. The statistics collection unit is also resource limited because of the vast number of nodes in the network to be queried.
There are two limitations that may limit the capacity in the network for statistics collection: system level limitations and per-node limitations. System level limitations include: (i) time required for building requests and managing responses, (ii) upper bound on the number of items that can be transmitted with a single request and (iii) upper bound on outstanding requests in a segment of the network given capacity of the links in the segment and the number of expected responses. Per-node limitations include: (i) node response time, (ii) node resource limitations and (iii) upper bound on the number of outstanding requests for the node given that the node must divert switching resources to respond to such requests. An outstanding request is a request for which a response from the node has not yet been communicated to the statistics collection unit.
In prior art systems, requests for statistics for a specific interval are sent to the network in blocks which are not modelled for bandwidth or other efficiencies. As such, the system and per-node constraints may combine to defer requests to a node so that statistics collection is not maximized for a given interval. This may result in a loss of statistical data. Timely polling of the counters in a node must occur or the overall statistics collection process will produce incomplete or inaccurate results.
Additionally, in the prior art, a statistics collection unit may use an overly conservative constraint for the number of outstanding requests allowed for a particular node. A typical conservative constraint used is the smallest number of outstanding requests allowed for the nodes in the segment or network as the constraint for every node in the segment or network. This may underutilize a given node and defer further requests for that node until the current outstanding requests are answered. Deferring requests to the node may result statistics collection not being maximized for a given interval resulting in a loss of statistical data.
There remains a need for a system and method which improves upon the use of available resources in processing requests for statistics to nodes in a network.
SUMMARY OF THE INVENTIONIn a first aspect of the invention, a method of defining an order for sending a plurality of requests for statistics to an associated plurality of nodes of a communication network is provided. One or more requests are associated with an individual node. Each node has one or more node attributes. The method includes defining a sequence for sending the requests to be sent to the nodes. The sequence is based on a value of a selected node attribute of each node. The method also includes initiating each of the requests according to the sequence.
The method may define for each node a translated value related to the value of the selected node attribute prior to defining the sequence and base the sequence on a ranking of all of the translated values.
The method may select two or more node attributes and for each of these node attributes, refine the ranking of the nodes based on the value of each of those node attributes.
The method may receive data regarding a plurality of responses from the nodes which received requests. The method may track a number of outstanding requests in a segment of the communication network. The segment includes the associated nodes. The number of outstanding requests relates to a number of requests in the segment for which responses have not been received. The number of outstanding requests is associated with an upper bound number. The method may also initiate each of the requests by comparing the number of outstanding requests in the segment with the upper bound number and initiate one request when the number of outstanding requests in the segment is less than the upper bound number.
The method may receive data regarding a plurality of responses from each node which received the requests. The method may track a number of outstanding requests for each node. The number of outstanding requests for each node relates to a number of requests for which responses have not been received for each node. Each number of outstanding requests is associated with an upper bound number. The method may, for each node, initiate the requests when the number of said outstanding requests is less than the upper bound number, each request being sent independently to its associated node.
The method may separately define the upper bound number for each node in the segment.
The method may select a selected node attribute which has a value representing a number of requests to be initiated for each node in a time interval. The method may define the sequence by ranking the nodes in descending order utilizing each of the values of the selected node attribute of the nodes.
The method may select a selected node attribute which is a response time of each of the nodes to previous requests. The method may define the sequence by ranking the nodes in descending order utilizing each of the values of the selected node attribute of the of nodes.
The method may select a selected node attribute which identifies an operating characteristic of each of the nodes. The method may define the sequence by ranking the nodes in a predetermined order utilizing each of the values of the selected node attribute of the nodes.
The method may utilize an operating characteristic which indicates wireless and non-wireless transmission technologies associated with each of the nodes. The method may define the sequence by ranking the nodes utilizing values of the operating characteristic and ranking nodes having wireless transmission technologies with a higher priority.
The method may be embodied in a computer program.
The method may be repeated in a cyclic time interval.
The method may track a second number of other outstanding requests for each of the nodes. The second number of other outstanding requests for each of the associated nodes relates to a number of requests for each of the nodes for which responses have not been received. Each second number of other outstanding requests is associated with a nodal upper bound number. The method may initiate each request when the number of outstanding requests for the segment is less than the upper bound number and the second number of other outstanding requests is less than the nodal upper bound number for the individual node associated with each request.
In a second aspect of the invention, a statistics collection unit associated with a communication network is provided. The communication network includes a plurality of nodes and each of the nodes has one or more node attributes. The statistics collection unit includes a computer and a program executed on the computer. The program includes a sequencing module defining a sequence for sending requests for statistics to be sent to an associated plurality of nodes. One or more requests are associated with an individual node. The sequence is based on a value of a selected node attribute of each node. The program also includes an initiating module initiating each of the requests according to the sequence.
In a third aspect of the invention, an apparatus for use in a statistics collection unit in a communication network is provided. The communication network includes a plurality of nodes. The apparatus includes a device defining a sequence for sending the requests for statistics to be sent from the statistics collection unit to an associated plurality of nodes in the segment.
In a fourth aspect of the invention, a computer executable program for use on a communication network is provided. The communication network includes a plurality of nodes. The computer executable program executes the step of defining a sequence for sending a plurality of requests for statistics to be sent to an associated plurality of nodes. One or more requests are associated with an individual node. Each node has one or more node attributes. The sequence is based on a value of a selected node attribute of each of the nodes. The computer executable program also executes the steps of initiating each of the requests according to the sequence.
In other aspects, the invention provides various combinations and subsets of the aspects described above.
BRIEF DESCRIPTION OF THE DRAWINGSThe foregoing and other aspects of the invention will become more apparent from the following description of specific embodiments thereof and the accompanying drawings which illustrate, by way of example only, the principles of the invention. In the drawings, where like elements feature like reference numerals (and wherein individual elements bear unique alphabetical suffixes):
The description which follows, and the embodiments described therein, are provided by way of illustration of an example, or examples, of particular embodiments of the principles of the present invention. These examples are provided for the purposes of explanation, and not limitation, of those principles and of the invention. In the description which follows, like parts are marked throughout the specification and the drawings with the same respective reference numerals.
The foregoing embodiment relates to a method and apparatus for processing requests for statistics to be sent to the nodes in the network in a defined sequence. The manner of defining the sequence in this embodiment is configurable. Therefore, the requests may be sent in an intelligent manner, utilizing information gathered from the network to define a sequence in which to initiate the requests, providing efficient use of the bandwidth resources of the connection to the network and of the nodes themselves. Alternatively, the sequence may be defined to minimize the number of selected resource tracking statistics which may be reset before their data is retrieved.
First, a description of a communications system associated with an embodiment is provided. Referring to
It will be appreciated that network cloud 110 may be any packet switched network. For example, network cloud 110 may utilize asynchronous transfer mode (ATM) protocol for communication. In an ATM network, nodes 161-171 may be ATM nodes or switches.
Nodes 161-171 in network cloud 110 are grouped into segments, or subnets, 112-114. First segment 112 includes nodes 161-164; second segment 113 includes nodes 165-168; and third segment 114 includes nodes 169-171.
End users 181-184 may be connected to network cloud 110 via routers 190 and 191, allowing data to be transmitted amongst end users 181-184 through nodes 161-171 in network cloud 110. For example, a packet may be sent from end user 184 to end user 181, from user 184, to router 190, to network cloud 110, to router 191 and finally to end user 181.
As is illustrated, each of node segments 112-114 operably communicates with its associated statistics collection unit 141-143. Statistics collection units 141-143 may communicate to node segments 112-114 via dedicated network management functionality communication links 154-156. Each statistics collection unit 141-143 is responsible for gathering the appropriate statistics within its associated node segment 112-114. Statistics collection units 141-143 typically use a general purpose computer with specialized network management software operating thereon which gathers statistics from their segments 112-114.
The following description of an embodiment will refer to statistics collection unit 143, link 156, segment 114, and nodes 169-171 to illustrate aspects of the embodiment within a segment of network cloud 110. It will be appreciated that the same general description will apply to statistics collection unit 141, link 154, segment 112, and nodes 161-164 and statistics collection unit 142, link 155, segment 113, and nodes 165-168.
As described earlier, traffic transmitted through network cloud 110 may be monitored using counters. As such, each of nodes 161-171 comprise a number of counters for storing a set of statistics relating to traffic flow through the nodes and possibly other characteristics of the node. Again, the counters may be embodied in hardware, software or other means known in the art. Typically, the counters are reset after a predetermined time interval regardless as to whether or not their data has been collected. If the data from the counters is not collected in each interval, some data will be lost and the overall statistics collection process will produce incomplete or inaccurate results.
Referring to
Referring to
The statistics collection configuration information contains information concerning the nodes and their subcomponents required to respond to network statistics configuration 122 which is used to determine the protocol, frequency of polling and other parameters needed for the requests. Statistics management block 120 sends each set of statistics collection configuration information to its associated statistics collection unit 141-143.
When statistics collection unit 143 receives statistics collection configuration information from statistic management block 120, statistics collection unit 143 will form and send statistics requests to nodes 169-171 in node segment 114. Each individual request from statistics collection unit 143 is associated with, i.e. addressed to, a node 169-171 in segment 114 from which it requests statistics. It will be appreciated that statistics requests sent to the various nodes in network cloud 110 may be sent using SNMP or other polling protocols. It will be appreciated that statistics collection unit 143 may use software and/or hardware to process the initiation and processing of requests to nodes 169-171 in its segment 114. As described previously, statistics collection unit 143 in the embodiment uses a general purpose computer with specialized network management software operating thereon. However, other embodiments may have hardware embodied in customized circuitry, an ASIC, a microcontroller, and other electronics. The software may be embodied in one or more computer programs executing on one or many pieces of the hardware.
Statistics are sent from nodes 169-171 to statistics collection unit 143 in response to scheduled polling and event driven collection. In scheduled polling, the statistics collection unit 143 periodically sends requests to nodes 169-171 to respond with the required data. The requests may be sent at regular or irregular intervals. In event driven collection, requests are not sent by statistics collection unit 143.
An example of the type of statistics that may be tracked is the number of data packets being sent through network cloud 110 from end user 181 to either of end users 183 and 184. Statistics management block 120 receives network statistic configuration 122 relating to this requirement and generates statistics collection configuration information. The statistics collection configuration information is passed to each statistics collection unit 141-143. In this example, the individual requests are sent to a number of nodes 161-171 in a number of segments 112-114 in network cloud 110. Each statistics collection unit 141-143 collects the statistics corresponding to traffic sent by end user 181 from its associated nodes 161-171 via sending individual requests to one or more of nodes 161-171. It will be appreciated that the sequence in which the individual requests are sent by statistics collection unit 143 may affect how much data is gathered from each of nodes 169-171 in each interval since both system and per-node limitations can defer the arrival of individual requests for statistics or the responses thereto.
As described earlier, there are limitations in processing requests. System level limitations include the time required for building a request, the number of items that can be transmitted with a single request and the upper bound on outstanding requests in segment 114 given the capacity of links in segment 114 and the number of expected responses from nodes 169-171. Per-node limitations include the response time for each of nodes 169-171, computational processing limitations for nodes 169-171 and the upper bound on the number of outstanding requests for each of nodes 169-171 that marks when each node must divert switching resources to respond to such requests. An outstanding request is a request for which a response from its associated node 161-171 has not yet been communicated to its statistics collection unit 141-143.
The embodiment utilizes algorithms to initiate requests in a sequence to make efficient use of available resources, including bandwidth, of the node and the network. Alternatively, the sequence can be defined so that the number of requests to specific counters (chosen by the network operator) is maximized. It will be appreciated that, in some circumstances, resource usage may be managed by defining the sequence for initiating requests using a time-response algorithm that allows response to a greater number of requests from all nodes. In other circumstances, resource usage may be managed by defining the sequence on a hierarchical node ordering system, collecting data from specific nodes first.
Referring to
The network operator selects a particular node attribute 312-316 to be the selected node attribute for segment 114 or network cloud 110. The network operator may base this selection on the past performance of nodes 169-171. Choices for the selected node attribute may include: (i) the number of requests sent to nodes 169-171 in an interval (node attribute 312), (ii) the average response time for receiving a response from nodes 169-171 (node attribute 313), (iii) the class of node associated with the request (node attribute 314), (iv) the number of timeouts per node (node attribute 315) and (v) the number of counters on nodes 169-171 from which to collect data (node attribute 316). It will be appreciated that node attributes are not limited to those identified above and may be any information accessible about the node, its past performance in the network, information available from the network, previously gathered statistics and other attributes that may be derived from this information.
Once a selected node attribute has been selected, statistics collection unit 143 defines a sequence in which to initiate statistics requests to nodes 169-171 within segment 114. Representations of nodes 169-171 in segment 114 are ordered by the value of their selected node attributes according to the predetermined ranking algorithm. In this way, the sequence is based on the value of the selected node attributes of nodes 169-171. The requests are then initiated in the sequence based on the ranking of the representation of their associated nodes 169-171. It will be appreciated that initiating the requests is subject to system and per-node constraints that may affect the sequence in which they are sent.
The algorithm used to define the sequence is executed on the hardware that comprises statistics collection unit 143. This hardware is part of the device which defines the sequence and initiates the requests within statistics collection unit 143. The device may be a central processing unit, a piece of hardware specially hardwired for this task, a microcontroller, or a general purpose computer, among other things. Alternatively, it will be appreciated that a device separated from statistics collection unit 143 may be provided with hardware and software to access sets of node attributes 369-371 and define a sequence for initiating the requests in segment 114. The device then communicates the sequence to its associated statistics collection unit 143 to initiate the requests based on that sequence, subject to system and per-node constraints that may affect the sequence of the requests.
Defining the sequence of requests is illustrated by the following example. First, the network operator selects node attribute 312 as the selected node attribute for segment 114 and wishes to define the sequence for initiating requests in ascending order of this value. Statistics collection unit 143 accesses sets of node attributes 369-371 for the value of node attribute 312 for nodes 169-171. The values for node attribute 312 are 9, 2 and 6 for nodes 169, 170 and 171 respectively. In this example, the algorithm defines the sequence as initiating requests to node 170, node 171 and node 169, in that order, subject to system and per-node constraints. Therefore, statistics collection unit 143 will initiate requests in this sequence subject to these constraints.
The network operator may select two or more node attributes 312-316 to be selected node attributes. When two or more node attributes 312-316 are selected as selected node attributes, the network operator also specifies the order in which the selected node attributes will be used to define the sequence. Selecting more than one selected node attribute results in a ranking that is iteratively refined where the first selected node attribute is used to rank the representation of its associated node and then representations with the same value or range of values of the selected node attribute for the first ranking are ranked by the next selected node attribute. The representations with the same values or ranges of values for the selected node attributes for all of the previous rankings are then ranked by the next selected node attribute. This continues until the representations have been ranked by all of the selected node attributes.
For example, network operator chooses node attributes 314 and 315 as selected node attributes. The values of node attributes 314 and 315 are to be ranked in ascending numerical and alphabetical order and node attribute 315 is to be used to rank the representations of nodes 169-171 before node attribute 314. The algorithm ranks the representations for nodes 169 and 171 before node 170 when ranking by node attribute 315. The algorithm then refines the ranking for each group of representations with the same value for previous node attributes, nodes 169 and 171 and node 170 by node attribute 314. This results in a sequence for initiating requests for node 171, node 169 and node 170, in that order.
Algorithms used to rank the representations of nodes 169-171 by the value of their selected node attributes may differ depending on the information in the selected node attribute and the method for defining the sequence for initiating the requests. For example, ranking by the value of node attribute 312 by statistics collection unit 143 may rank the representations in ascending or descending order, depending upon the desired result. If the selected node attribute contains a character string, as does node attribute 314, the algorithm may rank the by the value of the selected node attribute alphabetically. Alternatively, the algorithm may rank by the value of the selected node attributes in a predetermined order based on the possible values for the selected node attributes. For example, the algorithm may sequence class “B” attributes first, followed by class “C” and class “A”, in that order. It will be appreciated that an operating characteristic of the nodes may be this class distinction. An example of such an operating characteristic may be the distinction between nodes which correspond to wireless transmission technologies and nodes that do not. A network operator may cause the algorithm to rank the wireless class of nodes before other types of nodes thus giving a higher priority to the wireless class of nodes.
The algorithm used to define the sequence for initiating the requests may translate the selected node attributes into values that are more readily sortable prior to ranking them. The translated value for nodes 169-171 is a function of the value of their selected node attributes. The algorithm then uses the translated value to define the sequence for initiating the requests. For example, a selected node attribute containing a real number may be converted to an integer. Another example is an algorithm that translates node attribute 314 having a value of class “A” to the integer 1, class “B” to 2 and class “C” to 3 and ranking them in ascending order. The translation portion of the algorithm may also define the values for selected node attributes within a certain range to have the same value for their translated values. Whether a translation occurs depends on the type of information found in the selected node attribute and how the algorithm is used to define the sequence based on those values.
In the embodiment described, the algorithm ranks the representations of the nodes only in ascending order and maps selected node attribute values to an integer value used for this ranking. The desired sequence for initiating the requests will affect how this translation occurs. For example, referring to
In the embodiment, the requests themselves are not ranked but representations of the nodes with which they are associated are ranked. The ranked representations are used to determine the order in which nodes should be polled. The sequence of requests is defined by sending requests for the nodes in the order of this ranking. The association of requests with the representations of their nodes may be maintained using an array of pointers. It will be appreciated that the requests themselves may be stored in a sequenced order to accomplish the same goal of initiating the requests in a defined sequence. Also, it will be appreciated that it may not be necessary for the representations of nodes to be stored in a sequenced order prior to sending the requests to those nodes. This may be accomplished through an array of pointers to the representations of the nodes where the array of pointers indicates the sequence of nodes to which requests will be sent.
Referring to
One such constraint is the limit on the capacity of the links in segment 114, for example, the capacity of link 156 for statistics collection unit 143. Each link in segment 114 is able to receive only a predefined number of messages at a time, therefore, there is an upper bound on the number of outstanding requests in segment 114 to prevent the links from being overloaded with messages. It will be appreciated that the value of this upper bound may be different in each segment 112-114 and may be some number other than the actual upper bound.
To track statistics related to the upper bound in the segment, a counter is maintained by statistics collection unit 143 to count the number of outstanding requests in segment 114. The counter value is incremented for every request sent and decremented for each response received. Before a request is sent, the counter is compared with the segment upper bound. If the counter has not yet reached the segment upper bound, the request is sent. If the counter has reached the segment upper bound, the request is not sent. This ensures that the segment upper bound will not be exceeded before sending a request for segment 114. It will be appreciated that this counter may be embodied in hardware or software using techniques known in the art.
There may also be an upper bound on the number of outstanding requests that exist for each node 169-171. This upper bound exists for each node 169-171 because nodes 169-171 must divert resources away from their switching functions to respond to the requests for statistics. Exceeding this upper bound for nodes 169-171 may cause traffic interruptions and loss of service for the node. Therefore, this upper bound should not be exceeded. Statistics collection unit 143 maintains a separate counter for each node 169-171 in segment 114. The upper bound for each node 169-171 is compared with its counter in the same manner as the segment upper bound. It will be appreciated that these counters may also be embodied in hardware or software using techniques known in the art.
Constraints such as those noted above may cause statistics collection unit 143 to send requests in an order other than the sequence defined by the described method. Typically, the upper bounds for the number of outstanding requests for each segment 112-114 are much larger than the upper bound for the number of outstanding requests for any given node 161-171 in network cloud 110.
In the embodiment, the methods for ensuring the segment and the per-node upper bound for outstanding requests are not exceeded are both performed by statistics collection unit 143. It will be appreciated that, in other embodiments, statistics collection unit 143 may perform only one of the methods of checking the upper bounds or may not perform any method of checking the upper bounds.
When nodes 169-171 within node segment 114 respond to the outstanding sent statistics requests, the responses are collected by statistics collection unit 143. When statistics collection unit 143 sends requests to segment 114, it may do so through a standard interface known in the art, such as a socket. Statistics collection unit 143 stores appropriate structures to receive responses from each request. The structures are associated with the socket corresponding to its request. Each response arrives on the same socket on which it was sent. Statistic collection unit 143 combines the information from the responses with the other statistics received from segment 114 and stores this information in its corresponding structure. It provides these relevant statistics to statistics aggregation block 130.
It will be appreciated that this process of defining a sequence for sending requests will be repeated for each set of statistics collection configuration information sent to each statistics collection unit 141-143. This may occur on a cyclic time interval since statistics collection configuration information may be in response to scheduled polling from statistics management block 120.
Statistics collection unit 143 may also update the information stored in sets of node attributes 369-371 (
Referring to
Referring to
Referring to
At 0 seconds, using the sequencing algorithm described earlier, request 537 is sent for the statistics to node 407, therefore, no further requests may be sent for node 407 since the upper bound for outstanding requests for node 407 has been reached. Request 538 will not be sent until request 537 returns a response. Since request 538 will not be sent at 0 seconds, request 531 is sent for node 401. At this point, the upper bound for the number of outstanding requests in segment 415 has been reached and no further requests are sent. At 1 second, request 531 has received a response and new requests may be sent. Request 532 is sent at 1 second. Again the upper bound for the number of outstanding requests in segment 415 has been reached and no further requests are sent. At 2 seconds, request 532 has received a response and request 533 for node 403 is sent. At 3 seconds, responses have been received from requests 533 and 537. Request 538 for node 407 is then sent at 3 seconds, along with request 534 for node 404. This process continues with requests 535 and 536 being sent at 4 and 5 seconds respectively.
The timeline in
The elements of
In the example of
It will be appreciated that the device defining the sequence for initiating the requests may alternatively employ a self-teaching method of defining a sequence for initiating requests without having the network operator select a selected node attribute. The self-teaching method could define the sequence based on a number of pre-programmed factors to optimize the collection of statistics from certain nodes or all of the nodes in the segment or network. The pre-programmed factors may be based on the results from previous collections performed with different sequencing methods. The self-teaching method may be performed in a simulated environment and the optimal method recommended to network operators.
An additional feature of the embodiment described above is the use of individual upper bounds for the upper bound on the number of outstanding requests allowed for the nodes in the network.
Referring to
In prior art systems, statistics collection unit 606 typically uses an overly conservative constraint for the number of outstanding requests for a node, such as the smallest number for all nodes 601-603, as the constraint for each of nodes 601-603 in segment 609 or network 610. In the example presented, the smallest upper bound from column 604 is one outstanding request.
Referring to
Using the individual upper bounds for each node 601-603, the amount of bandwidth in link 608 used by each of nodes 601, 602 and 603 is now one outstanding request for each of nodes 601 and 602, respectively represented by shaded portions 621 and 622 and three outstanding requests for node 603, represented by shaded portion 623. This allows node 603 in
Using the method of defining a sequence for initiating requests described with reference to
It will also be appreciated that in another embodiment an algorithm may be used to define the sequence which is a hybrid of algorithms described herein.
It is noted that those skilled in the art will appreciate that various modifications of detail may be made to the embodiments described herein, which would come within the spirit and scope of the invention as defined in the following claims.
Claims
1. A method for collecting statistics from nodes of a communication network having a plurality of heterogeneous nodes, comprising:
- defining for the network a node level limitation pertinent to statistics collection;
- for each node involved in statistics collection, mapping a respective value of the node level limitation to a corresponding node identification; and,
- sending statistics requests to each node based on the respective value, wherein each node processes a number of requests according to the respective value.
2. The method of claim 1 wherein the node level limitation includes a maximum number of outstanding packets to be transmitted from each node.
3. The method of claim 1 wherein the node level limitation includes a maximum number of requests per second that can be processed by the node.
4. The method of claim 1 wherein the defining includes determining the respective value from node information retrieved from the network.
5. The method of claim 1 wherein the defining includes determining the respective value for each node based on network level limitations pertinent to statistics collection.
6. The method of claim 1 wherein each node includes a respective number of statistics collecting counters, each counter for maintaining statistics pertinent to a node attribute.
7. The method of claim 1 and further comprising, for maximizing collection of statistics over a predetermined time interval:
- defining for each node a set of attributes for which statistics need to be collected;
- mapping a set of attributes collectable for each node to the corresponding node identification and generating a table with the mapping;
- storing in the table for each node identification a default value for each attribute in the set of attributes based on past performance of each node;
- ordering the statistics requests in a sequence of statistics requests according to a predetermined ranking calculated from the table; and,
- pooling the nodes each time interval according to the sequence of statistics requests for updating the table with current values of the attributes.
8. The method of claim 7 wherein the predetermined ranking is re-calculated for each time interval.
9. The method of claim 7 wherein criteria for the predetermined ranking is one or more of: per node attributes, a number of subcomponents on a node, a number of requests per node, an average response time for a node, and a number of timeouts per node.
10. A system for collecting statistics from nodes of a communication network having a plurality of heterogeneous nodes, comprising:
- a computer coupled to the communication network; and,
- a program executed on the computer, the program including:
- a module for defining for the network a node level limitation pertinent to statistics collection;
- a module for, for each node involved in statistics collection, mapping a respective value of the node level limitation to a corresponding node identification; and,
- a module for sending statistics requests to each node based on the respective value,
- wherein each node processes a number of requests according to the respective value.
11. The system of claim 10 wherein the node level limitation includes a maximum number of outstanding packets to be transmitted from each node.
12. The system of claim 10 wherein the node level limitation includes a maximum number of requests per second that can be processed by the node.
13. The system of claim 10 wherein the module for defining includes a module for determining the respective value from node information retrieved from the network.
14. The system of claim 10 wherein the module for defining includes a module for determining the respective value for each node based on network level limitations pertinent to statistics collection.
15. The system of claim 10 wherein each node comprises a respective number of statistics collecting counters, each counter for maintaining statistics pertinent to a node attribute.
16. The system of claim 10 wherein the program further includes, for maximizing collection of statistics over a predetermined time interval:
- a module for defining for each node a set of attributes for which statistics need to be collected;
- a module for mapping a set of attributes collectable for each node to the corresponding node identification and generating a table with the mapping;
- a module for storing in the table for each node identification a default value for each attribute in the set of attributes based on past performance of each node;
- a module for ordering the statistics requests in a sequence of statistics requests according to a predetermined ranking calculated from the table; and,
- a module for pooling the nodes each time interval according to the sequence of statistics requests for updating the table with current values of the attributes.
17. The system of claim 16 wherein the predetermined ranking is re-calculated for each time interval.
18. The system of claim 16 wherein criteria for the predetermined ranking is one or more of: per node attributes, a number of subcomponents on a node, a number of requests per node, an average response time for a node, and a number of timeouts per node.
Type: Application
Filed: Jan 19, 2006
Publication Date: Jun 1, 2006
Inventors: Wlodek Olesinski (Nepean), Rashid Qureshi (Ottawa), Mark Woodward (Carp)
Application Number: 11/334,466
International Classification: H04J 1/16 (20060101);