CN100518079C - Distributed performance data acquisition method - Google Patents
Distributed performance data acquisition method Download PDFInfo
- Publication number
- CN100518079C CN100518079C CNB2004100811351A CN200410081135A CN100518079C CN 100518079 C CN100518079 C CN 100518079C CN B2004100811351 A CNB2004100811351 A CN B2004100811351A CN 200410081135 A CN200410081135 A CN 200410081135A CN 100518079 C CN100518079 C CN 100518079C
- Authority
- CN
- China
- Prior art keywords
- performance data
- task
- collection
- module
- distributed
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Images
Abstract
Disclosed a distributed property data gathering method, comprising: (a), the step that defining the type of property data to generate the type; (b), attaining the gathering object and the property of property data; (c), based on the step (a) and step (b), generating the property gathering mission according to the form of property data; (d), based on the step (c), realizing the transformation from the property gathering mission to the property data; (e). based on the step (c), distributing each module of property data gathering engine; (f), based on the step (d) and step (e), utilizing and storing the gathered property data into the data base or plane document; (g) displaying the property data via the image or the forms for reporting statistics to the managers. The invention via the distribution of gathering point of system can meet the demand of variable network scale, and support the study, analysis, and prediction on the flow quantity character and the application condition of device of managed network.
Description
Technical field
The present invention relates to performance data acquisition method, relate in particular to the distributed performance data acquisition method of each module execution that is applied to IP network do as one likes energy data capture engine.
Background technology
Performance data acquisition method belongs to based on the catenet of advanced networks to be measured and management demonstration system category, according to the FCAPS definition of ISO (International Organization for standardization ISO) to network management, a large-scale complex network is managed, the performance management that need carry out will be conceived to the collection of the performance data of equipment in the large-scale ip network, storage, analysis process.
The performance data that the network management personnel obtains by the performance management mode in the network management, to help the network management personnel by watch-dog flow and utilize situation, what occur in the discovery network is unusual, fault as equipment, equipment is attacked or virus is invaded etc., and existing network-management tool is not enough to the control and monitoring of network, generally only provides the monitoring to separate unit or small number of devices, and the performance data type of monitoring is not expanded flexibly yet.
Because the performance data type in the network is various, and various characteristics that are magnanimity along with device type quantity, therefore not only satisfy the demand of general webmaster, research to network flow characteristic, prediction, very big help is provided, and the proposition of the performance collection method that the continuous expansion of the scale of network also is well solved by the distribution performance of collection point in the acquisition system is imperative.
Summary of the invention
The present invention proposes in order to address the above problem just.Distributed performance data acquisition method of the present invention, each module that is applied to IP network do as one likes energy data capture engine is carried out, and comprises the steps:
A) step of the formal generation form of definition performance data;
B) utilize configuration data acquisition module TOPDISCOVER in the performance data capture engine to obtain the step of the acquisition target and the attribute thereof of performance data;
C) on the basis of step a) and step b), generate the step of the task configuration module MISSION_CONFIG generation performance acquisition tasks in the form through performance data capture engine according to the formalization of this performance data;
D) on the basis of step c), the distributed performance acquisition module DDCE in the through performance data capture engine obtains this performance collection task, and reads in memory queue, realizes the step of performance collection task to the conversion of performance data by multithreading;
E) on the basis of step c), the distributed coordination center module MANAGER in the do as one likes energy data capture engine distributes performance data collection task for each collection point according to task allocation algorithms, and each module of performance data collection engine is carried out the step of distributed coordination;
F) on the basis of step d) and step e), to the utilization of the performance data of gathering and store database into or flat file in step;
G) performance data is presented to administrative staff's step by the form of expression of figure or form.
According to the present invention, by the distributivity of collection point in the system, well adapted to the demand of the continuous variation of network size, also provide a great help simultaneously for traffic characteristic, the research of machine utilization, analysis, prediction to managed networks.
Description of drawings
Fig. 1 is according to distributed performance data acquisition method general flow chart of the present invention;
The postfix expression calculation flow chart that Fig. 2 obtains through the head sea orchid for the computing formula in the performance data collection task;
Fig. 3 is the automatic configuration flow figure of performance collection task;
Fig. 4 is the general structure block diagram of collection process;
Fig. 5 is the flow chart of the master control thread of collection process after initialization;
Fig. 6 is the initialized flow chart when gathering process initiation among Fig. 5;
Fig. 7 is the flow chart of thread function in the data acquisition process;
Fig. 8 is the collecting thread flow chart;
Fig. 9 is the acquisition system structured flowchart;
Figure 10 is the flow chart of the task allocation algorithms one of distributed coordination center module MANAGER enforcement;
Figure 11 is the flow chart of the task allocation algorithms two of distributed coordination center module MANAGER enforcement; And
Figure 12 is a control module CONTROLLER control flow chart;
Figure 13 is the conversion schematic diagram of performance collection task to performance data; And
How Figure 14 is interpreted as the block diagram of a performance data for task of explanation.
Embodiment
With reference to the accompanying drawings specific embodiments of the invention are described in detail.
In the IP network management, the classification of performance data comprises: response time, throughput, utilance, degree of communication, the autgmentability of network.Obtaining these performance datas is convenient to the network management personnel network is carried out present situation monitoring, fault location, and future anticipation.And these performance datas all in various degree can by the equipment mib information obtain the reflection.All meet the IP device of webmaster standard, all should: 1) provide and follow RFC (Request For Comments, the internet research and development is the community of purpose, in a series of document, set forth the description of many agreements and model, experimental conclusion and review, nearly all internet standard agreement has all been write as RFC.) MIB-II of the standard (node of the MIB of RFC1213 definition, this node has defined the protocol-dependent management information of all standard webmasters, MIB and SMI, snmp protocol is listed as three chief components of Simple Network Management Protocol (simple networkmanagement protocol) framework, MIB has defined the set of the management object that can conduct interviews by NMP, usually being taken as is the virtual data base of management object, all information of management object are organized with tree structure, wherein each information object node in the tree can be located with OID, the corresponding concrete example of each information object, value that each example is bound, snmp protocol, it is the application protocol that defines for network management services, be that NMS is unified and operated in the agreement that the Agent on the managed object carries out asynchronous request and response, network management system can be passed through and agency's exchange snmp protocol message, come the relevant information among the MIB on the managed object at inquiry proxy place, the SMI structure of management information, be information object definition and the coding among the MIB, so that the basis of protocol transmission); 2) the self-defining Proprietary MIB information of each manufacturer (user management for convenience of many manufacturers has added self-defining information in MIB, these information are generally under the PRIVATE node in MIB).
Based on SNMP, be stored as purpose at network supervisor for reaching performance data collection, the present invention proposes a general performance data acquisition method, be applicable to the IP network of various scales by the performance data collection engine of this method definition, the performance data collection main-process stream that the present invention realizes comprises the steps: a) the formal generation form of definition performance data as shown in Figure 1; B) utilize configuration data acquisition module TOPDISCOVER in the performance data capture engine to obtain the acquisition target and the attribute thereof of performance data; C) on the basis of step a) and step b), generate the performance acquisition tasks according to the task configuration module MISSION_CONFIG in the formalization generation form through performance data capture engine of this performance data; D) on the basis of step c), the distributed performance acquisition module DDCE in the through performance data capture engine obtains this performance collection task, and reads in memory queue, realizes the conversion of performance collection task to performance data by multithreading; E) on the basis of step c), the distributed coordination center module MANAGER in the do as one likes energy data capture engine distributes performance data collection task for each collection point according to task allocation algorithms, and each module of performance data collection engine is carried out distributed coordination; F) on the basis of step d) and step e), to the utilization of the performance data of gathering and store database into or flat file in; G) performance data is presented to administrative staff by the form of expression of figure or form.
Below as shown in fig. 1 distributed performance data gathered main-process stream be described in detail.
1.1 the formal generation form of definition performance data
1) formal definition performance data collection flow process is 2 a tuples＜acquisition target, set of acquisition parameters 〉.
2) corresponding again 4 tuples＜object number of acquisition target wherein, address, version, safety certification 〉.
The √ object number is the unique number of each monitored object in the whole network.
The √ address is the access mode of object of monitoring, promptly refers in IP network, and the capital equipment of facing is IP device all in the webmaster scope as main frame, switch, the IP address of router etc.
The snmp protocol version that the √ version is to use, SNMP is from V1 up to now, and V2C has developed into V3C, and new version has constantly added better protocol operation mode, and has strengthened fail safe.
The √ safety certification is the mode of guarantee agreement safe operation, usually the password field that is used as visiting with a SNMP COMMUNITY field among the SNMP specifies the value of this field will make the agency can differentiate whether the manager has the authority of visit MIB in snmp protocol now.
3) set of acquisition parameters has then defined concrete performance collection data according to different image data types, needed MIB gathers the OID point, with calculate the needed account form of final performance data by these OID points, also represent＜the OID group with the form of one 3 tuple, computing formula is returned and is deposited parameter 〉.
√ OID group is meant and obtains several corresponding on the needed MIB tree of performance data OID numberings.
The √ computing formula is meant that the value that is obtained by OID is calculated the mode of performance numerical value.
√ returns the mode of depositing and is meant performance data at database, the form of depositing in the media such as flat file.
1.2 obtain the acquisition target and the attribute thereof of performance data
Configuration management requirement according to network, network management system must be obtained the configuration data of the whole network institute watch-dog, these configuration datas are also as the basis of performance data collection with prepare, use snmp protocol can obtain needed configuration data from the MIB of acquisition target, following table is the configuration data that the specific performance data are relied on.
The storage of performance data for convenience, we are the object refinement and the numbering of all properties data target correspondence, below are a kind of mode classification of suggestion: subnet, the network segment, host port, main frame, switch ports themselves, switch, router port, router etc. are respectively it and work out a kind of data type
Subnet--the network segment
+--main frame
+--host port
+--CPU
+--disk
+--switch
+-switch ports themselves
+--CPU
+--the router-Router port
Each object type will be endowed a type number.All like this configuration datas just can store classifiedly according to data type.
1.2.1 obtain the acquisition target in the performance data collection task--TOPODISCOVER
TOPODISCOVER is the configuration data acquisition module, will obtain the attribute information of all managed objects in the network.
The acquisition target that √ finds will be given the whole network unique identifying number (OBJECT_INSTANCE) and be made up of prefix and sequence number, and prefix has identified the type of equipment, as router one 100 switches, 1200 grades and sequence number often increases progressively successively according to the order of Topology Discovery.
Therefore the router in network can be represented as
11002 etc., switch ports themselves 15001,15002.Network equipments configuration tables of data from the Topology Discovery system, a record of the equipment in OBJECT_INSTANCE location comprises the IP and the OID call number of equipment, if the parent device of this equipment has a lot of fraternal equipment identical with this device type, when gathering this device performance data, promptly distinguish these fraternal equipment with the OID index, such as gathering switch and all switch ports themselves information all in the MIB storehouse of switch, each port has an OID call number in the MIB storehouse of switch, identify the position among the MIB of this port information place.
√ SNMP version is taked highest version usually, to the most complete prepreerence mode of data support, so because SNMP V1 supports 32 bit data types at most, relate to IF-MIB (coming a MIB node of statistic flow with 64 digit counters) the OID SNMP V2C agreement of Gbit flow number, remaining task SNMP V1.
The READ-COMMUNITY of community of Agent SNMP AGENT on all managed objects of √ will sound out according to the tabulation that the user provides.
TOPODISCOVER scans continuous configuration information to network, obtain up-to-date configuration data in real time, in case have new equipment to add or deletion, to send form such as $object_instance:object_type:configure_type to the task configuration module: message, defined object number respectively, device type under the object and the action type that need carry out this equipment: the deletion task is still added task, because the additions and deletions of task can not in time be reflected to the performance data collection process, therefore must periodically notify all control modules, restart all collection processes by Consultation Center.
1.2.2 the parameter group in the performance data collection task--PROPERTY_OID
Parameter group in the performance data collection task will be stored in a database table PROPERTY_OID, each subparameter that each field in the form is exactly a set of acquisition parameters.
OID group in the √ performance data formalization definition, the managed object information of standard mib storage is quite complete, these information can obtain by the OID of inquiry MIB, these information can not really provide mostly a network equipment performance condition (such as a large amount of in the MIB storehouse be the statistics of certain facility information from the system initialization to the current time, administrative staff are more interested to be interior situation of change of a period of time, i.e. differences in two time points), need to the request of OID node to originally handle, add up, analyze, obtain performance data.
The computational methods of this performance data of utilance of reference device port are from RMON (the telemanagement node the MIB provides the statistical information in a lot of these RMON node monitoring local area network (LAN)s):
UTILIZATION＝((PACKETS*(96+64)+OCTECTS*(8))/INTERVAL*107)*100％
If this performance data that is directly provided from RMON then needs count PACKETS, the corresponding OID of byte number OCTECTS by request message, obtain the packet receiving of router port and receive initial data such as byte, and calculate by above formula.
Be integrated into the initial data row of a performance data, become the parameter of performance data collection task in OID sequence mode.OID sequence form such as OID1; OID2; OID3; OID4; .... this OID sequence is number to be obtained by the OID prefix sequence+port index of corresponding task among the task description table PROPERTY_OID.
Example: the OID prefix sequence that certain port I/O flow of router is gone into bag is:
1.3.6.1.2.1.2.2.1.11；1.3.6.1.2.1.2.2.1.12；
The index number of this port in the all-router port is x, just can obtain the OID sequence of actual energy usefulness:
1.3.6.1.2.1.2.2.1.11.x；1.3.6.1.2.1.2.2.1.12.x；
Computing formula in the √ performance data collection task, the formula of original data computation in the performance data collection task, take the method for expressing of head sea orchid,, be through the blue postfix expression that obtains that transforms of head sea to the formula of shape such as 5+ ((8-2) * 3-(2+2* (12-9)+3)):
5 8 2-3 * 2 2 1 2 9 - * + 3 + - +
The stack data structure mode has been used in result's calculating,
The example: 1+2*4-5 change into 124 * 5+-
Obtain result of calculation the number of inverse Polan expression and symbol pop down successively and according to postfix expression calculation process shown in Figure 2,
Example:
Computing formula and OID sequence are finished a performance data computing jointly.Computing formula is the blue form of the head sea of shape such as g2_v0 (operand) g1_v0+ (operator), and before the computing, operand can be replaced by the numerical value that OID obtains, and in the operand gn_vm symbol, n is the collection period number, and m represents to replace with m numerical value of OID sequence.
Example:
1) calculating switch ports themselves t1 to the computing formula that interior I/O of t2 time on average goes into to wrap number (number/second) is:
((the non-unicast packet of t2 unicast packet+t2)-(the non-unicast packet of t1 unicast packet+t1))/(t2-t1)
2) port receives that the OID prefix sequence of unicast packet and non-unicast packet is:
1.3.6.1.2.1.2.2.1.11；1.3.6.1.2.2.1.12；
3) add the call number x of this port;
1.3.6.1.2.1.2.2.1.11.x；1.3.6.1.2.2.1.12.x；
4) computing formula: ((g2_v0+g2_v1)-(g1_v0+g1_v1))/t;
5) be converted into the computing formula of storing in the task: g2_v0 g2_v1+g1_v0 g1_v1+-t;
Two time points are obtained the numerical value of OID respectively, period 1 is used 1.3.6.1.2.1.2.2.1.11.x, the numerical value of the unicast packet that collects substitutes the g1_v0 symbol, use 1.3.6.1.2.1.2.2.1.12.x, the numerical value of the non-unicast packet that collects substitutes the g1_v1 symbol, uses 1.3.6.1.2.1.2.2.1.11.x second round, value replace g2_v0,1.3.6.1.2.1.2.2.1.12.x replace g2_v1, if acquisition interval is T in the task parameters, the time interval T of twice collection oid just can replace t.
6) next according to the pop down computational algorithm, obtaining the result is exactly the value that performance data switch ports themselves I/O on average goes into to wrap number.
The √ data are returned the mode in deposit data storehouse, to SQL statement
Insert into tablename values (object_instance, value0, taskid sequence nextvalue) calculates when finishing, table_name, object_instance, value0, taskid replace to performance data storage table name, object number, performance data and task number are carried out the insert library operation.
√ frequency acquisition (INTERVAL) in the time interval of twice collection of task of two cycle differences, directly replaces the t symbol in computing formula.The frequency acquisition that defines in the general networking traffic monitoring program is 5 minutes, 10 minutes, and 15 minutes, the CPU that system provides on many network equipments, data such as LOAD are generally also to get the mean value in these time intervals.The data volume of such acquisition target generation every day is 60/5*24=288.
1.3 the formalization form according to performance data generates the performance acquisition tasks
Each field that we can define the acquisition tasks tabulation by each attribute of last joint performance data is＜ip, community, version, cyckept, interval, oids, formula, sql 〉, describe below how task attribute and acquisition target are assembled into these acquisition tasks.
1.3.1 performance data is abstract to the performance collection task
A performance data can be abstracted into a task, and being abstracted into of task has reached following purpose:
√ is convenient to the conversion to performance data
Taskization determined, task is to finish collection to a performance data at a time point, still finishes collection to a collection of performance data a time period.Consider the flexibility that simple and easy degree that message is transmitted and task are adjusted, as shown in figure 13, a task of native system is finished with the collection to a performance data target in time period of a kind of frequency.
The self-contained conversion of parameter in the task obtains the method for performance data, it is computing formula, an acquisition module gathering level is after obtaining task list, task parameters must comprise gathers total time, at interval, the device object of gathering, obtain initial data, the method for calculated performance data, the position of performance table data store and field format in the database.
Following table has been enumerated the parameter that obtains from task list:
| Field | Meaning | Example | |
| OBJECT_INSTANCE | The managed object identifier | 11004 | |
| IP | The IP address of object place equipment | 192.168.7.7 | |
| SNMP VERSION | The SNMP version that uses | 1 | |
| COMMUNITYSTRING | The SNMP community that uses | PUBLIC | |
| OID GROUP | The OID set of initial data | 1.3.6.1.2.1.1： 1.3.6.1.2.1.1： | |
| FORMULATION | The computing formula of initial data | g2_v1g2_v0+ g1_v1g1_v0+- | |
| SQLSTRING | Performance index are returned the mode in deposit data storehouse | INSERT INTO TABLE | |
| INTERVAL | The task frequency acquisition | 300 | |
| | Task number | 0 |
Figure 14 has illustrated how task is interpreted as performance data.
1.3.2 acquisition tasks configuration module--MISSION_CONFIG
Network manager combines acquisition target and performance data parameter by task configuration module MISSION_CONFIG, is assembled into the performance collection task of being carried out by the performance collection interpretation of programs.Automatically configuration feature has realized that the acquiescence task parameters is the batch configuration task of all devices all properties data, the manual configuration function has then added with gerentocratic mutual, and the manager uses manual configuration can be chosen as specific equipment or specific performance properties data configuration task.All will enable the task configuration feature when system is initial, the task of when therefore disposing being Task Distribution is female task, and task number is 0, and the collection process of carrying out female task is called female process.When disposing automatically, acquisition module will obtain the OBJECT_INSTANCE numbering that the acquisition tasks distribution module obtains equipment to be configured from whole network configuration data interface, configuration data with equipment, comprise IP and OID call number corresponding in the acquisition tasks, device type sign OBJECT_ID, can from internal system table PROPERTY_OID, inquire all task types of this equipment correspondence according to OBJECT_ID, and these task type correspondences the CYCLE (cycle) of task, OIDS (the OID sequence that this task need be gathered), FORMULA (formula), and SQL (returning the sql statement that use in the deposit data storehouse), COMMUNITY (SNMP community that give tacit consent to when adding automatic configuration or that manually set, default configuration is PUBLIC when disposing automatically), SNMP VERSION (SNMP version, default configuration is the SNMPV1 version when disposing automatically) INTERVAL (collection period, default configuration is 300seconds when disposing automatically) the LASTING TIME (duration, automatically be defaulted as infinite during configuration), promptly can be according to cartesian product:
The form of IP * OID_INDEX * CYCLE * OIDS * FORMULA * SOL * COMMUNITY * SNMPVERSION * INTERVAL is combined into the performance collection task.
Fig. 3 is the automatic configuration flow figure of performance collection task.Manual configuration promptly is the process that adds the inquiry user in the centre, and the user can shield own uninterested equipment, the configuration of the task of SNMP version or performance data type.When disposing most of performance collection task, the processing method of system is similar, the many CPU of main frame and many disks task are more special, these two kinds of task types are that operating system is relevant, not only an equipment can dispose a plurality of performance data collection tasks, and the OID and the computing formula of getting with the different operating system task of a kind of task type are also different.
1.4 the performance collection task is to the conversion of performance data
The DDCE module gets access to the performance collection task, and reads in memory queue, obtains task by multithreading from formation, by the conversion of the realization of the flow process among Fig. 4 from the performance collection task to performance data.The distributing process of N acquisition module of the acquisition layer of the system level bottom, control module may be controlled some processes wherein on the wherein same harvester.Accept the start-up parameter of control module when gathering process initiation and make initial work, the master control thread is done the work of treatment of the reading of performance data collection task, control messages, promptly obtain allocating task and put into the performance data task stack, start collecting thread then the performance collection task is interpreted as performance data from database.
The task number of performance collection task, configuration module is given the task number (TASKID) of Task Distribution in configuration task be 0, therefore most task is that TASKID is female task of 0 in the task storage list, when new task is added into or revises, the Coordination module of whole system can distribute the TASKID that do not have in the task list (enchashment have maximum adds among all TASKID) to give this task, and the collection process just can obtain the task that configuration module is made by task number from database.
1.4.1 data acquisition process--DDCE
Fig. 5 is the flow chart of the master control thread of collection process after initialization.
The master control thread main flow process of collection process after initialization comprises the steps: that the parameter of the process of resolving is (frequency acquisition is gathered total time for task number, task type); Read in task stack and calculate required collecting thread number according to task number, task type according to number of tasks; According to frequency acquisition TIMER is set, total time periodicity is set according to gathering; Start the corresponding collecting thread function of number of threads and deposit thread, begin to gather with returning; Judge whether to receive message; If the message of receiving receives that WM_TIMER was through with to one-period, the master control thread wakes all collecting threads up; Go back to the step that judges whether to receive message; If the message of not receiving is returned and to be deposited thread and take out the result go back to the deposit data storehouse from result queue; 1 does the database cleaning work when accepting close message that control module sends, and 2 notice control modules are resumed work accordingly; Withdraw from the collection process.
The function package of acquisition layer is in executable file (COLLECTOR.EXE), and it is the collection process that the direct control module of cooperation layer starts in the mode of SHELL (operating system passes a kind of mode of ginseng to process) parameter.
Accept the parameter of the control module SHELL transmission of the direct control module of cooperation layer when acquisition layer is gathered process initiation, parameter format is:
√ task number TASK_ID.
The task type MISSION_TYPE1 that √ place acquisition control module is assigned to, MISSION_TYPE2---＞MISSION_TYPE N.
√ frequency acquisition INTERVAL.
√ gathers total time LASTING TIME.
Owing to used TIMER mechanism, need in the master control thread, create forms, the forms handle is called in as the parameter of START and RUN method, and these forms also will be accepted the message of bolt down procedure simultaneously.The name form of creating forms is TASK+A (TASKID), and for example female process creation forms then call createwindow (task0).
Fig. 6 is the initialized flow chart when gathering process initiation among Fig. 5.
Initialization flow process when gathering process initiation comprises the steps: to be provided with the step of database connection string; The step of the sql statement that the setting operation record sheet is used; The step that connects database; The step of closing database; The step of Thread Count is set; The master control thread reads the step of task stack; The forms handle of master control thread creation is imported into the step of collecting thread; Start collecting thread and enter the step that periodic message is handled.
OO encapsulation, class missionmachine is the encapsulation to the performance collection task, does not become as long as configuration task database table form, this class just can well be reused, and has especially expanded collection total time, frequency acquisition and collecting thread Several Parameters.Make that the customization of performance data collection task is more flexible.
1.4.2 the thread function in the collection process
Fig. 7 is the expansion of the thread function in the data acquisition module flow chart.
The flow process that thread function is launched is as follows: begin to judge that whether task stack is not for empty; If be not empty, then from task stack, get task; Fetch the corresponding initial data (get_oid_group) of oid from managed object; Calculate performance data (computefunction) according to formula; Performance data is filled the result data structure, and put into formation; If task stack is empty, then thread oneself is hung up oneself.If above-mentioned formula is monocyclic computing formula, then directly calculate; If be double-periodic computing formula, calculate according to the collection result of twice of front and back, has calculated afterwards with secondary collection result alternative before once.
Topmost two methods of class missionmachine are FillAttri () and collecting thread function f nThread ().
Among the method Fillattr of √ class missionmachine, read the performance data task from database and put into task stack missionqueue (result that the stack of a STL stl deposits is task mission).We use storehouse to deposit task, be to consider if the task queue of FIFO, the task that SNMP response is fast more will constantly shift to an earlier date in formation so, responding slow or overtime more task will not have no progeny and move, form conditions of streaking, the time span that makes formation finish like this is more and more longer, finally may surpass collection period, therefore we use first-in last-out storehouse constantly to fall task back and forth, make finishing more soon of task in preceding top of storehouse and bottom, and can be constantly to intermediate compression, this effect of actual conditions is fine.
Among the √ method start, collecting thread function f nThread and result return and deposit thread storagethread and will be activated, and by missionqueue.pop (), a plurality of fnThread thread function are fought for task from formation, carry out the explanation of task then.Gather function f nthread and use the SNMP++ storehouse, this storehouse has been carried out OO encapsulation to all message structures of SNMP.The flow process of function is as follows:
1) resolves from formation, ejecting of task, obtain the IP address of acquisition target, COMMUNITY and OID sequence.
2) carry out the REQUEST operation of SNMP with the get_oid_group function.
The main process of get_oid_group as shown in Figure 8.
3) so far the oid sequence results has been kept in the vb array (vb is the numerical value binding, and a vb can regard one＜oid name, oid value as〉to), calls that computing function ComputeFunction calculates and checkout result insertion result queue.When computing function ComputeFunction calculated two periodic duties, the value of last one-period collection is saved and the collection value in this cycle is calculated together, and after calculating was finished, the numerical value that this cycle collects, preservation was used for following one-period.Calculate the main blue formula of head sea that adopts and calculate the computing formula of formula correspondence among the mission with stack architecture, specific algorithm exists " parameter group in the performance data collection task " discuss in the joint, repeat no more.
4) performance data is calculated finish back and database and is returned and deposit sql composition result structure, is pressed into resultqueue (queue of result, this is the formation of a FIFO); Return to deposit and eject the result structure in the thread, requltqueue.pop (), the performance data among the result is gone back to the deposit data storehouse in the mode of sql appointment.Specifically in joint expansion down.
1.4.3 returning of performance collection task result deposited
Return and to deposit thread function storagethread and constantly resultqueue is inquired about.
√ formation non-NULL then takes out the result from formation, replaced among the sql respective symbol and finished the insert library operation by the performance data value in the result structure and current time.
The √ formation is empty, before the next cycle acquisition tasks first time is not finished, does not have the result and inserts result queue at least, for preventing the inquiry meaningless to result queue, the sleep certain hour.
The task that we adopt Fair Queue to deposit back to deposit in the actual conditions, that is to say, after each collecting thread is finished the work, a queue number that increases progressively rollback will be obtained successively, deposit result queue in according to this formation, return and deposit and a thread of a formation correspondence therefrom obtains task, this mechanism makes all task stacks can keep length even, has guaranteed to greatest extent that existing Thread Count can be finished back to deposit in the shortest time.
1.5 the distributed coordination of each module of performance data collection engine
Below with reference to Fig. 9 the distributed coordination of each module of performance data collection engine is described.
TOPODISCOVER，MISSION_CONFIG，DDCE，MANAGER，CONTROLLER，DATAPROCESSOR，MISSIONBROWSER
TOPODISCOVER wherein, MISSION_CONFIG, DDCE are the nucleus module of performance collection system, the configuration data acquisition module, the task configuration, distributed performance data acquisition module, front had been introduced the function of these modules, comprised the collection of configuration data, the configuration of acquisition tasks, poll with all devices in the net territory that this module is responsible for is interpreted as performance data to the performance collection task, and deposits database in.
MANAGER is the distributed coordination center, is responsible for the configuration of performance collection task, the Task Distribution between each DDCE, startup and the end of control DDCE
CONTROLLER is a control module, is responsible for transmitting the communication information between MANAGER and the DDCE.
MISSIONBROWSER is for being embedded into the task browse module of other network management systems, and the user can implement browsing of performance acquisition tasks by this module, adds and modification.
The distribution mode of acquisition tasks, can guarantee the fault-tolerance of system, thereby flow separation is avoided gathering flow backbone equipment is caused interference between each management domain, but the coordination between each module must solve by mutual communication, the message from interface layer is monitored by Consultation Center, realizes concrete task coordinate function.
The embodiment of system's distribution character:
The √ transparency, concerning acquisition system, as long as control module starts and registration, can accept the management of Consultation Center, the condition that control module must satisfy is, can deposit address information in database, Consultation Center can communicate by letter with control module by this address, and control module can search acquisition module in file system.
The √ fault-tolerance, mention among the TOPODISCOVER, Consultation Center can periodically start the collection process again, if at this moment behind Consultation Center and the control module communication failure, there are not image data or nearest acquisition time to substantially exceed collection period from now if inquire about the table at the place of the performance data type that this module is responsible for gathering, think that this equipment has broken down, calling task allocation algorithm so again, and then restart the collection process of each collecting unit.
√ efficient under the prerequisite of 5 clock collection period, is gathered under the CPU disposal ability condition of limited of main frame, and in order to satisfy the ever-increasing needs of network size, as long as constantly add new collection point, the registration control module just can be shared the task amount that falls to increase newly.
1.5.1 task allocation algorithms--the MANAGER that implement at the distributed coordination center
MANAGER is responsible for task by on all collection point of being assigned to of justice, the algorithm that distributes must be considered the weight of computer process ability, the pairing weight of time length that may spend with every kind of performance collection task, because configuration data can and be configured the cycle that data acquisition module scans according to the situation of network reality and constantly change, so MANAGER will redistribute task according to the interval certain hour cycle.Allocation algorithm will be rescheduled.Task, is distributed according to CONTROLLER according to the numbering in autonomous territory then by at first.
Large-scale network management system in one, the acquisition tasks amount tends to reach 100,000 grades in the collection period, and the performance data task type also has tens, discuss below all task distribution of cooperation layer task configuration module to harvester time, the principle that should follow, and provided the target function of planning:
Principle: identical performance data task type is assigned on the collection point, the task definition that the collection point is distributed is its resource that has, then need do to give a definition, the task number of certain task type and the product of task interpretation time have determined the relative size of this resource, and average resource is counted divided by collection for all resource summations.
Computer is born number of tasks according to computing capability, and the standard of computing capability can be CPU, memory size.
Target function: the resource allocation of supposing every computer is RESOURCEi, and the number of tasks of bearing is TASKi, and the general assignment number is TOTALTASK, and number of tasks and computing capability that every computer is born are suitable:
OBJECT＝MIN(TASKi-TOTALTASK*RESOURCEi/(RESOURCE1+….RESOURCEi))
Arthmetic statement
The computing capability of supposing all collection points all is the same, and such as identical CPU and memory source are arranged, principle is each collection point, and as the harvester of task, the computational load of its use is identical.
Figure 10 is the flow chart of task allocation algorithms one.Below with reference to Figure 10 algorithm one is described:
I task type arranged, and the number of each task is Ni, and be Ti the computing time of each task, and the number of collection point is M, and then each collection point Resources allocation averageresource obtains the resource totalresource that overall need calculate.
In the algorithm with certain particle, be the collection point Resources allocation, when resource surpasses average resource, then finish distribution to it, begin the distribution of next collection point, consider the compatibility of task in the real system, promptly work as a harvester because fault stops to gather the collection that only can have influence on specific several types task, it is certain task type that granularity delimited, be incremental number promptly being harvester task of distributing with a kind of number of task type, when the task of distributing surpasses the average of all acquisition tasks of system, continue the distribution of next harvester task.
The acquisition tasks number of algorithm simplification real system is many, the ratio that accounts in the general assignment number of the number of each task is less comparatively speaking, therefore granularity is smaller, (perhaps number of tasks is a lot of to be not easy to occur a very big task of consumption calculations resource, perhaps computing time is very long) situation that a supervisor is filled up, the time of Practical Calculation, time-delay+the CPU of SNMP communication carries out the time of formula operation, there is small difference the time of formula operation, the difference of the time of SNMP communication is more small, therefore can regard Ti equally as.
Figure 11 is the flow chart of task allocation algorithms two.Below with reference to Figure 11 task allocation algorithms two is described.
Algorithm one is when harvester is many, and average energy holds number of resources hour, then when the task number of some task type is many, causes the too much unbalanced situation of indivedual harvester number of tasks easily.Use complicated height of time this moment but the algorithm two of better effects if:
Earlier task is arranged from big to small according to resource size, remaining space is defined as average resource and deducts the number of resources that this collection point has been distributed, get the collection point of remaining space maximum, carry out resource allocation, because number of tasks restrains, each distribution all is best, therefore can totally reach best, real system has promptly adopted this allocation strategy, can reach comparatively ideal result.
The migration and the co-ordination sequential of the task that the distributed coordination center MANAGER of distributed performance data acquisition engine carries out comprise following work: the distributed coordination center can periodically start the collection process again, the distributed coordination center is judged when break down in certain collection point, must call described task allocation algorithms again so, and then the notice collection point, restart distributed performance acquisition module TOPDISCOVER by control module CONTROLLER.
Add task browse module MISSIONBROWSER task at the distributed coordination center and the request of revising handles accordingly.
1.5.2 task browse device--MISSIONBROWSER
The increase task, when the increase task operating appears at two kinds, a kind of is before system enters the initial launch state, must finish automatic configuration and manual configuration by the task configuration module, the interpolation task number is female task of 0, so that system's energy normal operation is a kind of in service in system, when the manager adds particular characteristic data acquisition task for particular device.Can inquire about the corresponding control module of these requests after Consultation Center accepts request, transmitting these tasks increases request, and control module starts corresponding collection process then.
The modification task, when the modification task occurs in system's operation, when the manager finds that the task parameters of some original configuration tasks is such as frequency, the operation total time of non-female task, SNMP version etc., in the time of need being changed, they will propose the request of change task by client so, and the corresponding control module of these requests can be inquired about by Consultation Center, transmit these tasks and change request, during the control module processing messages,, allow female task dormancy if this performance data has only female process gathering, carry out new task simultaneously, if the existing new task of a task is being carried out, the task deletion original substitutes with new task.When new task executed his total cycle, the process at the new task place notice that will finish to its control module transmission task made female process of corresponding dormancy be waken up.
1.5.3 control module starts flow process--CONTROLLER
Figure 12 is the control flow chart of control module.
Control module CONTROLLER is obtaining this machine IP at first, create common memory section, and insert this IP value as identifying the sign that control module has been activated, thereby to guarantee that a main frame can only move a control module, module is waited for the user input slogan, and the preface idol of IP and port composition is registered in database.
In database, search the contact method of notice Consultation Center, notify it that new control module registration and adding acquisition system are arranged.
Monitor the coordination request of Consultation Center, accept the message of Consultation Center, analytic message obtains request type, to increase, modification or the alarm frequency conversion task requests that occurs in the system, gathers the startup of process accordingly, reboot operation.
When this collection point need log off, then the user directly closes this control module, earlier this acquisition module is done to nullify in database when module is accepted to withdraw from request and operates.
1.5.4 performance data processing module--PROCESSOR
This module is responsible for the transfer to performance data, preliminary treatment, and provide suitable feedback for performance collection.
The √ performance data can be stored in according to the mode of secondary storage in the database, and the form of one-level storage is less, and the data of only storing nearest period of performance are convenient to user search, and secondary table is all historical informations, and table size is linear growth.The performance data processing module must be responsible for the transfer work of information, to the cleaning and the integration work of historical information,
The integration algorithm is as follows:
1) based on the method for time domain
Data sequence to a performance data.a1，a2，a3，a4，a5...
If define a coefficient step coefficient and be b (a2-a1)/a1＜b includes a2 in house object slightly so, calculate (a3-a1)/a1 then, if (an-a1) a1＞b so an substitute the position of a1, calculate (an+1-an) an＞b, our step function that obtains will be than originally lack a large amount of step points like this, and still big step still can be hunted down.
According to the definition of performance data, can be given up slightly data to all and average and obtain average discharge in the whole time period, more can reflect actual traffic conditions.
2) based on the method for frequency domain
Filter high-frequency signal with a low frequency filter.Original a1, a2, a3, a4, a5...
Frequencies filtration with 10 minutes obtains a1, a3, and a5 ..., obtain a1, a4, a7... with 15 minutes frequencies filtration
Said method can cause the loss of information, and according to the definition of performance data, we can obtain original a1 with a1+a2/2, interior and interior average discharge of a2 time period.Replace a1 and two numerical value in the time period of a2 place with a performance numerical value.
The preliminary treatment of √ performance data, performance data can be through a filter in shifting the way, this filter definition certain rule, performance data has surpassed the rule of filter definition, do as one likes can send a warning message by data processing module.Administrative staff can define different alarm levels and corresponding logical condition, mate according to rank mode from high to low, then transmission alarm that the match is successful,
For example: following table is the alarm conditions matching order
| Alarm level 1 A:a1＜x＜a2 |
| Alarm level 2 B:b1＜x＜b2 |
| Alarm level 3 C:c1＜x＜c2 |
If performance numerical value is positioned in the middle of the b1-b2, and do not belong to the scope of a1-a2, will send 2 grades of alarms so.In actual webmaster demand, need equipment is carried out the alarm that is provided with of batch, perhaps all equipment all is provided with alarm, the priority of other alarm individual equipment of so same level should be greater than batch setting greater than all the type equipment, in reality realizes, condition to batch device can be mapped as group's operation that individual equipment is carried out
Therefore the order of coupling is:
| |
| |
| Alarm level 2 B0:b1＜x＜b2 individual equipment |
| Alarm level 2 B1:b5＜x＜b6 all |
| Alarm level |
| 3 C0:c1＜x＜c2 individual |
| Alarm level |
| 3 C1:c5＜x＜c6 all devices |
Last table is for adding the matching order that all devices is provided with alarm conditions.
The frequency conversion of √ performance alarm is the special modification task requests that has added specific feedback mechanism.Performance data will be taken out by the performance alarm system, compare with the threshold values of setting according to certain rule, if find to have unusual, this unusual in order to verify, the performance alarm system can notify the performance data acquisition system, increases frequency acquisition, makes warning system more responsive to the seizure of performance data.
For example the numerical value of average discharge (PACKET PER SECOND is obtained divided by collection period by the difference that front and back relatively go into to wrap number for twice) in one-period of going into to wrap of certain router port reaches 0 value, warning system capture this unusual after, wish thisly unusually preferably to be got rid of by intensive collection at once, require to strengthen frequency acquisition, warning system foreshortens to 30s as collection period from default cycles 300s, so can send the frequency conversion alarm notification that router port goes into to wrap the flow task to the cooperation layer of acquisition system.It is similar that frequency conversion alarm and task are revised request, adds feedback and be because warning system when the frequency conversion task finishes, need return a internal task that warning system provides and number confirms in order to discern the frequency conversion task.
Getting in touch of direct control module of acquisition layer and acquisition module is the closest.Directly the control module function is encapsulated among the executable file CONTROLLER.EXE, can only move an acquisition controlling process on the collection main frame, from another angle, because acquisition control module directly is responsible for acquisition layer is gathered calling of process, the computational resource of a collection main frame also is responsible for allocation schedule by it.
1.6 the utilization of performance data and storage.
No matter performance data is stored in the database, or in the flat file, all must follow a certain form, and helps like this providing interface to other system, also is convenient to data are carried out further analyzing and processing.
The list structure of every performance data storage list can be following mode:＜performance data corresponding objects numbering, numerical value, acquisition time 〉.And the implication of every performance table can following XML file form interface externally is provided, then the DTD form of performance data storage list is:
＜! ELEMENT performance data storage list situation (the description * of table) 〉
＜! The description of ELEMENT table (device type, performance data is described, performance data table table name) 〉
＜! ELEMENT device type (#PCDATA) 〉
＜! ELEMENT performance data field description (#PCDATA) 〉
＜! ELEMENT performance data table table name (#PCDATA) 〉
1.7 the performance of performance data
Performance data can be presented to the network management personnel by figure and two kinds of manifestation modes of form.Wherein form can be done certain regular to initial performance data, such as doing one day, one week, one month average discharge, the average discharge of an equipment total interface, the average discharges of a subnet all devices etc. by curve tracing, can be observed the trend of certain device object performance numerical value change in cycle a period of time intuitively.
According to distributed performance data acquisition method of the present invention, its distributed characteristic makes acquisition system have the transparency, fault-tolerance, high efficiency advantage, distributivity by collection point in the system, the demand that has well adapted to the continuous variation of network size also provides a great help for traffic characteristic, the research of machine utilization, analysis, prediction to managed networks simultaneously.
More than distributed performance data acquisition method of the present invention is described in detail, the present invention is not limited to the foregoing description, those skilled in the art can make multiple modification or change in the process that the present invention is examined or puts into practice obviously, and the present invention is intended to cover these modifications or change.
Claims (10)
1, a kind of distributed performance data acquisition method is applicable to the IP network of various scales by the performance data collection engine of this method definition, and it is characterized in that: the step by each module of this performance data collection engine is carried out comprises:
A) step of the formal generation form of definition performance data;
B) utilize configuration data acquisition module TOPDISCOVER in the performance data capture engine to obtain the step of the acquisition target and the attribute thereof of performance data;
C) on the basis of step a) and step b), generate the step of performance data collection task according to the task configuration module MISSION_CONFIG in the formal generation form through performance data capture engine of this performance data;
D) on the basis of step c), the distributed performance data acquisition module DDCE in the through performance data capture engine obtains this performance data collection task, and reads in task stack, realizes the step of performance data collection task to the conversion of performance data by multithreading;
E) on the basis of step c), the distributed coordination center module MANAGER in the do as one likes energy data capture engine distributes performance data collection task for each collection point according to task allocation algorithms, and each module of performance data collection engine is carried out the step of distributed coordination;
F) on the basis of step d) and step e), to the utilization of the performance data of gathering and store database into or flat file in step;
G) performance data is presented to administrative staff's step by the form of expression of figure or form.
2, distributed performance data acquisition method as claimed in claim 1 is characterized in that:
Described performance data collection engine comprises with lower module: distributed coordination center module MANAGER, do unified performance Task Distribution and migration, and concentrate the work schedule of coordinating other modules; Distributed performance data acquisition module DDCE is used for the collection of performance data; Control module CONTROLLER is responsible for transmitting the communication information between distributed coordination center module MANAGER and the distributed performance data acquisition module DDCE; Task browse module MISSIONBROWSER, the user can implement browsing of performance acquisition tasks by this task browse module, adds and deletion; Configuration data acquisition module TOPDISCOVER obtains the configuration data of all managed objects in the network; Task configuration module MISSION_CONFIG combines acquisition target and performance data parameter, is assembled into the performance data collection task of being carried out by the performance collection interpretation of programs; Performance data processing module PROCESSOR is responsible for the transfer to performance data, preliminary treatment, and provide suitable feedback for performance data collection; Database module DATABASE is used for memory property data and performance task.
3, distributed performance data acquisition method as claimed in claim 1 is characterized in that:
The formal generation form of performance data is defined as two a tuples＜acquisition target in the described step a), set of acquisition parameters 〉, the corresponding again 4 tuples＜object number of described acquisition target, the address, version, safety certification 〉, described set of acquisition parameters is with 3 tuples＜OID group, computing formula is returned and to be deposited parameter〉form represent that wherein said OID is the sign of a management object of expression among the management information bank MIB.
4, distributed performance data acquisition method as claimed in claim 1 is characterized in that:
In the described step c), comprise automatic configuration feature and manual configuration function, automatically configuration feature can be long-pending according to Di Kaer by task configuration module MISSION_CONFIG: the form of IP * OID_INDEX * CYCLE * OIDS * FORMULA * SQL * COMMUNITY * SNMPVERSION * INTERVAL, according to corresponding performance data to device attribute, the different requirements of acquisition parameter, automatically be combined into the performance collection task, automatically allocating default is mixed the acquisition tasks of all correlated performance data automatically to all managed objects of the whole network, manual configuration is with regard to each attribute in the long-pending form of the above-mentioned Di Kaer of task, adding is mutual with user's inquiry, allows the user to adjust the value of these attributes;
Abbreviation during above-mentioned Di Kaer is long-pending is defined as follows:
IP is the IP address of acquisition target in network;
OID_INDEX is the OID call number, and described OID is the sign of a management object of expression among the management information bank MIB;
CYCLE is a duty cycle;
The OID sequence that OIDS need gather for this task;
FORMULA is a calculated performance data computing formula;
The SQL statement that SQL uses when depositing in database;
COMMUNITY is a SNMP community, and default configuration is PUBLIC when disposing automatically;
SNMPVERSION is the snmp protocol version that acquisition target is supported;
INTERVAL is a collection period.
5, distributed performance data acquisition method as claimed in claim 1 is characterized in that:
In step d), the N that the bottom of system level is distributing performance data acquisition module when moving as the process of operating system, done following work,
In the time of the process initiation of acquisition module, accept the start-up parameter of control module and make initial work,
The master control thread is done the work of treatment of the reading of performance data collection task, control messages,
Collecting thread is made the interpretation work of performance data collection task to performance data.
6, distributed performance data acquisition method as claimed in claim 5 is characterized in that,
Initialization flow process in the time of the process initiation of described acquisition module comprises the steps:
Frequency acquisition and the step of collection total time are set;
The step of database connection string is set;
The step of the sql statement that the setting operation record sheet is used;
The step that connects database;
The master control thread reads the step of task stack;
The step of closing database;
The step of Thread Count is set;
The forms handle of master control thread creation is imported into the step of collecting thread;
Start collecting thread and enter the step of periodically gathering.
7, distributed performance data acquisition method as claimed in claim 5 is characterized in that,
The flow process that described collecting thread launches is as follows:
Begin to judge whether task stack is empty;
If be not empty, then from task stack, get task;
Fetch the corresponding initial data of OID according to task description from managed object, wherein said OID is the sign of a management object of expression among the management information bank MIB;
Calculate performance data according to formula;
Collecting thread is as a result of put into result queue to performance data;
If task stack is empty, then collecting thread oneself is hung up oneself;
If above-mentioned formula is monocyclic computing formula, then directly calculate; If be double-periodic computing formula, calculate according to the collection result of twice of front and back, has calculated afterwards with secondary collection result alternative before once.
8, distributed performance data acquisition method as claimed in claim 2 is characterized in that:
Described distributed coordination center module MANAGER defers to task allocation algorithms task is assigned on all collection points liberally, the principle of following is: identical performance data task type is assigned on the collection point, the task definition that the collection point is distributed is its resource that has, then need do to give a definition, the task number of certain task type and the product of task interpretation time have determined the relative size of the resource of this certain task type correspondence, average resource is counted divided by collection for all resource summations, and the allocation algorithm of being taked is following algorithm one or algorithm two:
The resource that algorithm one, collection point are assigned with surpasses the average resource that it should obtain, and then finishes the distribution to it, begins the distribution of next collection point;
Algorithm two is arranged dissimilar tasks from big to small according to the resource size of their correspondences, and remaining space is defined as average resource and deducts the number of resources that certain collection point has been distributed, gets the collection point of remaining space maximum, carries out resource allocation.
9, distributed performance data acquisition method as claimed in claim 8 is characterized in that:
The migration and the co-ordination sequential of the task that the distributed coordination center module MANAGER of performance data collection engine carries out comprise following work:
The distributed coordination center module can periodically start the collection process again, the distributed coordination center module is judged when break down in certain collection point, so again, calling task allocation algorithm, and then the notice collection point, restart configuration data acquisition module TOPDISCOVER by control module CONTROLLER;
The distributed coordination center module is added task browse module MISSIONBROWSER task and the request of revising handles accordingly.
10, distributed performance data acquisition method as claimed in claim 2 is characterized in that:
Performance data processing module PROCESSOR does the preliminary treatment of performance data, comprise cleaning and integration work to historical information, do the performance data preliminary treatment, performance data is shifting filter of passage in transit, this filter definition certain rule, performance data has surpassed the rule of filter definition, and do as one likes can send a warning message by data processing module.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2004100811351A CN100518079C (en) | 2004-09-30 | 2004-09-30 | Distributed performance data acquisition method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2004100811351A CN100518079C (en) | 2004-09-30 | 2004-09-30 | Distributed performance data acquisition method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1756190A CN1756190A (en) | 2006-04-05 |
| CN100518079C true CN100518079C (en) | 2009-07-22 |
Family
ID=36689156
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2004100811351A Expired - Fee Related CN100518079C (en) | 2004-09-30 | 2004-09-30 | Distributed performance data acquisition method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100518079C (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106936660A (en) * | 2015-12-31 | 2017-07-07 | 华为软件技术有限公司 | Collecting method and device |
Families Citing this family (43)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101166108B (en) * | 2006-11-15 | 2011-06-01 | 南京大学 | Method for a distributed task dispatching soft bus with dynamic scalability |
| CN101370024B (en) * | 2007-08-15 | 2012-10-31 | 北京灵图软件技术有限公司 | Distributed information collection method and system |
| CN102105842B (en) * | 2008-07-24 | 2013-10-16 | Abb研究有限公司 | System and method of business rule integration with engineering applications |
| CN102033882B (en) * | 2009-09-25 | 2013-07-03 | 中兴通讯股份有限公司 | Method and system for storing performance data |
| US9665620B2 (en) | 2010-01-15 | 2017-05-30 | Ab Initio Technology Llc | Managing data queries |
| CN101895428B (en) * | 2010-07-27 | 2014-07-02 | 中兴通讯股份有限公司 | Acquisition system and acquisition point distribution method |
| CN101902364A (en) * | 2010-08-26 | 2010-12-01 | 中兴通讯股份有限公司 | System and method for reporting comprehensive network management data |
| CN102402459B (en) * | 2010-09-10 | 2013-09-11 | 中兴通讯股份有限公司 | Method and device for summarizing performance data of network management system |
| CN101989928A (en) * | 2010-11-05 | 2011-03-23 | 中兴通讯股份有限公司 | Network equipment information acquisition method and system, management system and network equipment |
| CN102682043B (en) * | 2011-04-14 | 2013-11-27 | 天脉聚源(北京)传媒科技有限公司 | Method for connecting distributed databases |
| US9116955B2 (en) * | 2011-05-02 | 2015-08-25 | Ab Initio Technology Llc | Managing data queries |
| CN102363814B (en) * | 2011-11-23 | 2013-08-28 | 山西农业大学 | Absolute quantification polymerase chain reaction (PCR) detection method |
| CN103198165B (en) * | 2012-01-05 | 2016-01-13 | 北京泛华恒兴科技有限公司 | Based on collecting method and the data collector of flexible test technology |
| CN102611576B (en) * | 2012-02-29 | 2015-02-04 | 大唐移动通信设备有限公司 | Method and device for generating network management configuration model file |
| CN103384206B (en) * | 2012-05-02 | 2016-05-25 | 中国科学院计算机网络信息中心 | A kind of method for parallel processing towards mass data and system |
| CN103384205B (en) * | 2012-05-02 | 2016-05-25 | 中国科学院计算机网络信息中心 | A kind of mass alarm data parallel acquisition system, device and method |
| CN103678726B (en) * | 2012-09-05 | 2017-11-10 | 亿阳信通股份有限公司 | A kind of data complementary collection method and system |
| CN103095492A (en) * | 2012-12-28 | 2013-05-08 | 中兴通讯股份有限公司 | Data collection method and data collection device |
| CN103209189A (en) * | 2013-04-22 | 2013-07-17 | 哈尔滨工业大学深圳研究生院 | Distributed file system-based mobile cloud storage safety access control method |
| CN104219071B (en) * | 2013-05-30 | 2017-12-15 | 华为技术有限公司 | The monitoring method and server of a kind of network quality |
| CN104426900B (en) * | 2013-09-11 | 2019-12-06 | 腾讯科技（深圳）有限公司 | multimedia data acquisition method and system |
| CN103716190B (en) * | 2013-12-25 | 2017-12-22 | 南威软件股份有限公司 | The method of E-government Intranet service application safety supervision |
| EP3108366A1 (en) * | 2014-02-18 | 2016-12-28 | Koninklijke Philips N.V. | Efficient processing of device related log files |
| CN103957124A (en) * | 2014-05-05 | 2014-07-30 | 上海北塔软件股份有限公司 | Distribution type collecting method and system of collection tasks |
| CN104331322B (en) * | 2014-10-24 | 2018-03-13 | 华为技术有限公司 | A kind of process migration method and apparatus |
| US10417281B2 (en) | 2015-02-18 | 2019-09-17 | Ab Initio Technology Llc | Querying a data source on a network |
| CN104881436B (en) * | 2015-05-04 | 2019-04-05 | 中国南方电网有限责任公司 | A kind of electric power communication device method for analyzing performance and device based on big data |
| CN105162628B (en) * | 2015-08-24 | 2018-12-18 | 上海天旦网络科技发展有限公司 | Quickly determine the system and method for the reasonable alarm threshold of network |
| CN106101207B (en) * | 2016-06-03 | 2019-07-12 | 东软集团股份有限公司 | Collector control method and device |
| CN106301926A (en) * | 2016-08-19 | 2017-01-04 | 北京北信源软件股份有限公司 | A kind of method shown based on WEB environment machine room lower various dimensions and monitor in real time |
| CN106533842B (en) * | 2016-12-20 | 2023-07-04 | 长沙先导智慧城市投资有限公司 | Partner type network supervision method and supervision equipment capable of being analyzed independently |
| CN107070748A (en) * | 2017-04-13 | 2017-08-18 | 周发辉 | A kind of processing system and method for the big data that communicates |
| CN109150930B (en) * | 2017-06-15 | 2020-12-18 | 北京金山云网络技术有限公司 | Configuration information loading method and device and service processing method and device |
| CN109993275B (en) * | 2017-12-29 | 2021-01-29 | 华为技术有限公司 | Signal processing method and device |
| CN108536539B (en) * | 2018-04-26 | 2021-12-17 | 东北大学 | Task scheduling method in industrial distributed data acquisition system |
| US11093223B2 (en) | 2019-07-18 | 2021-08-17 | Ab Initio Technology Llc | Automatically converting a program written in a procedural programming language into a dataflow graph and related systems and methods |
| CN110708208B (en) * | 2019-08-26 | 2023-01-06 | 奇安信科技集团股份有限公司 | Monitoring data acquisition method and device, storage medium and terminal |
| CN111177221B (en) * | 2019-12-26 | 2021-05-04 | 苏州亿歌网络科技有限公司 | Statistical data acquisition method, device and equipment |
| CN111475505B (en) * | 2020-03-27 | 2023-05-26 | 平安科技（深圳）有限公司 | Data acquisition method and device |
| CN111651644A (en) * | 2020-05-29 | 2020-09-11 | 山东中创软件商用中间件股份有限公司 | Method, system and device for identifying equipment type |
| CN113098734B (en) * | 2021-03-31 | 2022-07-08 | 科来网络技术股份有限公司 | Method for dynamically monitoring performance index of network equipment |
| CN113036930B (en) * | 2021-05-07 | 2022-07-12 | 国网福建省电力有限公司 | Power grid data communication management system |
| CN115277260B (en) * | 2022-09-28 | 2022-12-30 | 北京中超伟业信息安全技术股份有限公司 | Method and system for detecting vulnerability of cloud platform of Internet of things |
-
2004
- 2004-09-30 CN CNB2004100811351A patent/CN100518079C/en not_active Expired - Fee Related
Non-Patent Citations (4)
| Title |
|---|
| 基于SNMP网络性能管理的研究与实现. 赵金铃.中国优秀博硕士学位论文全文数据库. 2002 |
| 基于SNMP网络性能管理的研究与实现. 赵金铃.中国优秀博硕士学位论文全文数据库. 2002 * |
| 基于分布式主动代理的网络性能监测和发布系统. 程立，汪为农，李承.计算机工程，第27卷第2期. 2001 |
| 基于分布式主动代理的网络性能监测和发布系统. 程立，汪为农，李承.计算机工程，第27卷第2期. 2001 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106936660A (en) * | 2015-12-31 | 2017-07-07 | 华为软件技术有限公司 | Collecting method and device |
| CN106936660B (en) * | 2015-12-31 | 2020-02-14 | 华为软件技术有限公司 | Data acquisition method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1756190A (en) | 2006-04-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN100518079C (en) | Distributed performance data acquisition method | |
| CN104639374B (en) | A kind of application deployment management system | |
| CN101567814B (en) | Automatic network management method based on SNMP and stochastic Petri net | |
| US6393386B1 (en) | Dynamic modeling of complex networks and prediction of impacts of faults therein | |
| CN104598425B (en) | A kind of general multiprocessing parallel calculation method and system | |
| US7606895B1 (en) | Method and apparatus for collecting network performance data | |
| CN100521628C (en) | Expandable dynamic network monitor system and its monitor method | |
| CN106533806B (en) | Method for providing cross-layer QoS based on application awareness in multi-tenant SDN network | |
| US20030084156A1 (en) | Method and framework for generating an optimized deployment of software applications in a distributed computing environment using layered model descriptions of services and servers | |
| US5966509A (en) | Network management device | |
| US20030084155A1 (en) | Representing capacities and demands in a layered computing environment using normalized values | |
| CN102158360A (en) | Network fault self-diagnosis method based on causal relationship positioning of time factors | |
| JPH11212937A (en) | Production of report using distributed work station | |
| CN103795575B (en) | A kind of system monitoring method towards multiple data centers | |
| CN114302103A (en) | Intelligent monitoring operation and maintenance system | |
| CN114244676A (en) | Intelligent IT integrated gateway system | |
| JP2023506239A (en) | Systems and methods for autonomous monitoring and recovery in hybrid energy management | |
| US11652703B2 (en) | Dynamic processing distribution for utility communication networks | |
| CN108121639A (en) | A kind of data center's total management system design method based on cloud platform | |
| CN102045186B (en) | Event analysis method and system | |
| Chae et al. | Exposing the network: support for topology-sensitive applications | |
| CN110798660A (en) | Integrated operation and maintenance system based on cloud federal audio and video fusion platform | |
| EP1118952A2 (en) | System, method and computer program product for providing a remote support service | |
| CN109120443A (en) | A kind of management method and device of network attached storage NAS device | |
| Li et al. | Machine learning empowered intelligent data center networking: A survey |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20090722 Termination date: 20160930 |