CN100356729C - Method and system for monitoring network service performance - Google Patents

Abstract

The present invention discloses a method and a system for monitoring the performance of network services. In the method, parameters are analyzed by a baseline configured by a network management system so as to determine the range of service data and the type of the baseline; the network management system obtains the average value, the maximum value and/or the minimum value of the service data from the service data according to configured sampling service data so that one or a plurality of baselines reflecting the performance of the service are formed by using the average value, the maximum value and/or the minimum value as a reference point; the current service data is obtained from a network, the service data of the current sampling point is compared with the baselines, and when the service data values of continuous a plurality of sampling points are offset from one baseline at the same side of the baseline, the performance of the network services is determined to have a fixed variation tendency.

Description

Method and system for monitoring network service performance

technical field

The invention relates to communication network technology, in particular to a method and system for monitoring network service performance.

Background technique

In the communication network, operators need to grasp the operation status of the network in each period, and judge whether the operation has deteriorated or abnormal trend by analyzing the collected performance data. basis for tracking.

At present, in network performance analysis, especially in traffic analysis, operators only obtain relevant performance data, and make statistics according to the time axis and management object set. They can only obtain statistical information at a certain time, and cannot obtain the long-term operation trend or periodicity of the network. running trend.

Contents of the invention

The purpose of the present invention is to provide a method and system for monitoring network service performance, so as to solve the problem in the prior art that it is impossible to monitor the change trend of network service performance for a long time or periodically.

Realize technical method of the present invention:

A method for monitoring network service performance, the network has a network management system capable of collecting network service data, the method comprising the steps of:

Configure the baseline analysis parameters through the network management system. If it is a simple baseline, determine the baseline type and the number of historical data points;

Sampling business data from the network according to the configuration;

If it is a simple baseline, obtain the average value, maximum value and/or minimum value of historical data points from the business data, and use this value as a benchmark to form one or more horizontal baselines reflecting business performance, if it is a cycle If there is a permanent baseline, the average value, maximum value and/or minimum value of the basic unit of analysis of each distribution point in the cycle is obtained from the business data according to the number of historical data cycles, and a benchmark reflecting the business performance is formed based on this value. or multiple baselines;

Using the baseline as a reference, obtain the current service data value from the network and compare it with the value corresponding to the current time point on the baseline. When the value is , it is determined that there is a fixed trend in the service performance of the network.

When it is determined that the service performance of the network has a fixed change trend, an alarm message is generated or/and a pre-configured control strategy is executed to dynamically optimize the control of the network.

The baseline is a simple baseline representing relatively stable characteristics of traffic data within a period of time, and the baseline analysis parameters include baseline type, historical data point data, abnormal data elimination method and confidence interval.

The baseline is a performance periodic baseline that indicates that business data has a periodic change trend within a certain period of time. The baseline analysis parameters include cycle, historical data cycle number, basic unit of distribution analysis points, abnormal data elimination method, and confidence interval.

A network service performance monitoring system, comprising:

A performance database module for storing historical performance data;

A baseline database module for storing baseline configuration data and baseline data;

A business data acquisition module, used to collect actual data related to performance business indicators from the telecommunication network, and store them in the performance database module;

The baseline analysis configuration module is used to receive customized baseline analysis configuration parameters. If it is a simple baseline, determine the baseline type and the number of historical data points, and store the configuration data in the baseline database module. If it is a periodic baseline, determine the period type, history The basic unit of data cycle number and distribution point analysis, and store the configuration data in the baseline database module;

The baseline data analysis module reads historical performance data and baseline analysis configuration parameters from the performance database module and the baseline database module to form a baseline, and if it is a simple baseline, according to the average value, maximum value and/or The minimum value forms one or more horizontal baselines reflecting business performance for the benchmark point and saves them in the baseline database. If it is a periodic baseline, analyze the average value and maximum value of the basic unit according to each distribution point in the obtained cycle and/or the minimum value as a reference point to form one or more baselines reflecting business performance, and save them in the baseline database;

The network optimization analysis module is used to judge the quality of the network performance according to the baseline data, the service configuration of the network equipment and the collected current performance data, and perform corresponding optimization operations on the network.

The present invention has the following beneficial effects:

1. It can monitor the abnormal traffic in the telecommunications network, automatically locate and isolate the abnormal ports, especially suitable for controlling viruses and black Internet cafes.

2. By providing basic data of business performance changes, it is convenient for operation and maintenance personnel to perform network optimization, network planning, fault diagnosis, virus source discovery, and black Internet cafe tracking.

3. Eliminate abnormal data through the confidence interval to avoid the large deviation caused by the small number of sample points when performing baseline analysis and prediction.

4. Through the baseline interval method, an acceptable business performance change interval is given to the user.

5. The trend of performance change (worse/better) can be found early by baseline comparison of continuous and consistent change trends.

Description of drawings

Figure 1 is a schematic diagram of a telecom metropolitan area network;

Fig. 2 is a schematic structural diagram of the network optimization baseline analysis system of the present invention;

Fig. 3A is a simple baseline schematic diagram of the uplink port of the second regional office;

Fig. 3B is a simple baseline schematic diagram of the uplink port of the eight regional offices;

Figure 4A is a comparison chart between the current business data and the baseline of the second regional bureau;

Figure 4B is a comparison chart between the current business data and the baseline of the eight regional bureaus;

Figure 5 is a schematic diagram of the daily periodic baseline of the uplink port of the second regional office;

Fig. 6 is a schematic diagram of the comparison between the actual service data of the uplink port of the second regional office and the daily periodic baseline.

Detailed ways

Embodiment one

This example illustrates a simple baseline analysis of the performance of the present invention. Referring to the network of the telecom metropolitan area network shown in Figure 1, in such a hierarchically converged broadband metropolitan access network, it is divided into two regions, the second district and the eighth district, both of which use MA5 100 to provide ADSL service access, and all traffic After converging to ISN8850, it enters the metropolitan area network for transmission. Run the program on the network management system to collect traffic data. The scope of collection is the daily average rate of all ISN8850 uplink ports.

Referring to Figure 2, the network service performance monitoring system includes:

Performance Database: Holds historical performance data.

Baseline Database: Holds baseline configuration data and baseline data.

Business data collection module: responsible for collecting the actual data of relevant performance business indicators from the telecommunication network and saving them in the performance database.

Baseline analysis configuration template: responsible for receiving user-defined baseline analysis configuration parameters, and saving configuration data to the database for baseline analysis.

Baseline data analysis module: read historical performance data and baseline analysis configuration parameters from the performance database module and baseline database module respectively, form the baseline required by the user based on the baseline analysis method, and save it in the baseline database.

Network optimization analysis module: responsible for baseline comparison based on the baseline data, combined with the business configuration of network equipment and the collected current performance data, judging the quality of the network performance, outputting the network optimization analysis report, and performing automatic or manual optimization operations on the network configuration .

Performance simple baseline analysis includes: simple baseline analysis configuration, abnormal data elimination, simple baseline data analysis, and baseline comparison.

1. Simple baseline analysis configuration: To define and establish the historical data range of the baseline, the following related parameters need to be configured:

Baseline type: daily baseline, weekly baseline, monthly baseline, yearly baseline.

Number of historical data points: The number of historical data points required for simple baseline data analysis. The more the number of points is set, the more it reflects the long-term trend, and the less the number of points can well reflect the recent changing trend.

Abnormal data elimination method: eliminate abnormal data; do not eliminate abnormal data. All data will be used for baseline analysis without weeding out outlier data.

Confidence interval: Eliminate abnormal data using [average value - average value × n%, average value + average value × n%], where n% is the confidence level, and the value of n is 30, and the control of n can be realized by setting.

The simple baseline analysis configuration determines the scope of the raw data and the type of baseline. For the method of eliminating abnormal data, data preprocessing is required. Instead proceed to Step 3: Simple Baseline Data Distribution Analysis. Let's take the establishment of the daily baseline as an example to illustrate.

For example: refer to the historical data of the past 7 days to establish a simple daily baseline, and use a 30% confidence level to eliminate abnormal data. The parameters are:

Baseline Type: Daily Baseline

Number of historical data points: 7

Abnormal data elimination method: Eliminate abnormal data

Confidence interval: [mean - mean x 30%, mean + mean x 30%]

The raw data analyzed are as follows (seven days):

The actual rate of the ISN 8850 uplink port in Zone 2 (Gb/s)

1 2 3 3 4 5 6 7

0.7 0.45 0.65 0.5 0.60 0.55 0.85

The actual speed of the ISN 8850 uplink port in the eight districts (Gb/s)

1 2 3 4 5 6 7

0.8 0.69 0.9 0.76 0.42 0.85 0.94

2. Eliminate abnormal data

Abnormal data refers to data that deviates too much from the mean of all original data. Due to the small amount of data in the sample points, the calculation of the normal distribution is not applicable, and a simple confidence interval [mean value - mean value × n%, mean value + mean value × n%] is used to eliminate abnormal data. This initial average differs from subsequent baseline averages. For the data in the example we calculate as follows:

The actual rate of the ISN 8850 uplink port in Zone 2 (Gb/s)

Average value = (0.7+0.45+0.65+0.5+0.60+0.55+0.85)/7=0.614

Confidence interval: [0.614×0.7, 0.614*1.3]=[0.43, 0.80]

The 7th data 0.85 was eliminated.

The actual speed of the ISN 8850 uplink port in the eight districts (Gb/s)

Average value = (0.8+0.69+0.9+0.76+0.42+0.85+0.94)/7=0.766

Confidence interval: [0.766×0.7, 0.766*1.3]=[0.54, 0.99]

The 5th data 0.42 was eliminated.

3. Simple baseline data analysis

Simple baseline data analysis is to calculate the average value, maximum value, and minimum value of the effective data of historical data sampling points, thus forming three points, extending horizontally along the x-axis direction to form a baseline. For the data in the example, the calculated data is as follows, and the extension forms the daily baseline.

The actual speed of the ISN 8850 uplink port in Zone 2 (Gb/s):

Maximum value = MAX[0.7, 0.45, 0.65, 0.5, 0.60, 0.55] = 0.7

Min = MIN[0.7, 0.45, 0.65, 0.5, 0.60, 0.55] = 0.45

Average = AVG[0.7, 0.45, 0.65, 0.5, 0.60, 0.55] = 0.575

A simple baseline of performance in Zone 2 is shown in Figure 3A.

The actual speed of the ISN 8850 uplink port in the eight districts (Gb/s):

Maximum value = MAX[0.8, 0.69, 0.9, 0.76, 0.85, 0.94] = 0.94

Min = MIN[0.8, 0.69, 0.9, 0.76, 0.85, 0.94] = 0.69

Average = AVG[0.8, 0.69, 0.9, 0.76, 0.85, 0.94] = 0.82

A simple baseline of performance across the eight regions is shown in Figure 3B.

4. Baseline comparison

The network optimization analysis module compares the current performance data obtained from the network with the baseline data, and the comparison adopts a baseline interval diagram. Based on the current sampling point exceeding the baseline interval, we can understand the current business performance changes.

The rate data currently collected by the ISN8850 uplink port in Zone 2:

0.6 0.58 0.65 0.49 0.52 0.56 0.65

The comparison between the current data and the baseline data in Region 2 is shown in Figure 4A.

The rate data currently collected by the ISN8850 uplink port in the eighth area:

0.96 0.98 0.95 0.92 0.8

The comparison of current data and baseline data in the eight districts is shown in Figure 4B.

In comparison and judgment, referring to the quality sampling analysis method in the quality management system, generally adopting deviations that exceed the same direction for 7 consecutive times, it can be determined that the system has a fixed trend of change. The system judges whether the network performance has undergone a performance change in a certain trend by exceeding a certain baseline interval in the same direction several times in a row. The second number is determined by different business fields. Telecom networks generally use 4-7 points to calculate, and can also be adjusted (enlarged or reduced) according to the current actual business situation.

5. When it is determined that there is a fixed change trend in the service performance of the network, the pre-configured control strategy is executed to dynamically optimize the control of the network.

When taking 3 as the number of quality sampling deviations to judge the performance change trend. From the above two figures, it can be seen that the actual traffic of the current uplink port of the aggregation device in the second zone is within the baseline range, and there will be no major changes for the time being, and there is no need for capacity expansion. The actual traffic of the current uplink port of the aggregation device in the eighth area Three consecutive baseline intervals exceeding the maximum and minimum values of the baseline have occurred, indicating that the network traffic in this area is too large, and capacity expansion measures are urgently needed. In the actual implementation process, an access point in the eighth section is cut to the ISN 8850 in the second section, so as to realize network optimization without expansion investment.

Embodiment two:

This embodiment describes the periodic performance baseline analysis of the present invention. Also take the actual port rate of the ISN8850 uplink port of the second regional office in the network shown in Figure 1 at 8:00-9:00 AM every day as an example. The network service performance monitoring system is shown in Figure 2.

Performance periodic baseline analysis includes: periodic baseline analysis configuration, abnormal data elimination, periodic baseline data analysis, and baseline comparison.

1. Periodic baseline analysis configuration includes:

Cycle: According to business characteristics, the general cycle includes daily cycle, weekly cycle, monthly cycle and annual cycle.

Number of historical data cycles: The original data for baseline analysis is historical data of multiple cycles. The more historical data, the better it can reflect the general trend. Baselines will not reflect recent business characteristics well. Too little historical data is likely to cause large data deviations and large individual influences, which cannot be used as an effective performance baseline. A suitable cycle number of historical data is 3-8 cycles.

The basic unit of distribution analysis point: it is the unit of time point for distribution analysis within the cycle. Generally, the hour is used as the basic unit of distribution analysis points in the daily cycle; the day is used as the basic unit of distribution analysis points in the weekly cycle; the day is used as the basic unit of distribution analysis points in the monthly cycle; the month is used as the basic unit of distribution analysis points in the annual cycle.

Abnormal data elimination method: eliminate abnormal data; do not eliminate abnormal data. All data will be used for baseline analysis without weeding out outlier data.

Confidence interval: Eliminate abnormal data using [average value - average value × n%, average value + average value × n%], where n% is the confidence level, n is 30, and the control of n can be realized by setting.

The periodic baseline analysis configuration determines the scope of raw data and the type of baseline. For the method of eliminating abnormal data, data preprocessing is required. Instead go to Step 3: Periodic Baseline Data Distribution Analysis. The following is an example of establishing a daily baseline:

For example: the daily baseline is established by referring to the historical data of the port actual rate in the past 7 days, and the analysis points are distributed 24 hours a day. A 30% confidence level is used to eliminate abnormal data, and the parameters are:

Cycle: Daily cycle

Number of historical data cycles: 7

Distribution point analysis base unit: hour

Abnormal data elimination method: Eliminate abnormal data

Confidence interval: [mean - mean x 30%, mean + mean x 30%]

Take the 7-day simulated raw data of one of the distribution analysis points 8:00-9:00AM, as follows:

Unit: Gb/s

1 2 3 4 4 5 6 7

0.66 0.68 0.72 0.61 0.70 0.79 0.64

In order to establish a periodic daily baseline, the following processing needs to be performed for each distribution analysis point. For the case in the example, it is necessary to perform the following steps 2-3 for the 24-hour segment.

2. Eliminate abnormal data

Abnormal data refers to data that deviates too much from the mean of all original data. Due to the small sample data of the distribution analysis points, the calculation of normal distribution is not applicable, and the simple confidence interval [average-average×n%, average+average×n%] method (n=30) is used to eliminate abnormal data. This initial average differs from subsequent baseline averages. The calculation for the data in the example is as follows:

Average value = (0.66+0.68+0.95+0.61+0.70+0.79+0.64)/7=0.72

Confidence interval: [0.72×0.7, 0.72*1.3]=[0.50, 0.94]

The third data 0.95 was eliminated.

3. Periodic baseline data analysis

Periodic baseline data analysis is to use valid data to calculate the average, maximum and minimum values of each distribution analysis point in the cycle, so that multiple distribution analysis points in the cycle form three lines as a periodic baseline.

For the data of the above distribution analysis points, the calculation is as follows:

Maximum value = MAX[0.66, 0.68, 0.61, 0.70, 0.79, 0.64] = 0.79

Min = MIN[0.66, 0.68, 0.61, 0.70, 0.79, 0.64] = 0.61

Average = AVG[0.66, 0.68, 0.61, 0.70, 0.79, 0.64] = 0.68

The following steps 2-3 are performed on the 24-hour period respectively, and the three baselines formed by a total of 24 distribution analysis points are shown in Figure 5.

4. Baseline comparison

The performance data of the current cycle is obtained from the network, and these data are compared with the current baseline data. The comparison adopts the method of curve interval diagram, as shown in Figure 6. Based on the fact that a distribution analysis point exceeds the curve interval, the business performance change of a specific distribution analysis point can be understood.

Referring to the quality sampling analysis method in the quality management system, it is generally adopted that there are 7 consecutive deviations exceeding the same direction, and it can be determined that the system has a fixed trend of change. The system judges whether the network performance has undergone a performance change in a certain trend according to the multiple consecutive times in the period that exceed the baseline interval in the same direction. The number of consecutive times is determined by the distribution points in the period, and is calculated by using 15%-30% of the number of distribution points.

For the data output by simple performance baseline analysis and performance cycle baseline analysis, network optimization can be performed to ensure the service quality of the telecommunication network. For example, if the network interface traffic has a constant increasing trend, the system calculates how many days it will take to reach the maximum capacity of the current configuration of the network based on the current rate of change. When the network interface traffic reaches 60% of the maximum capacity, the system sends an alarm; when the network interface traffic reaches 80% of the maximum capacity, the system issues a serious alarm and forces the user to decide whether to increase the PVC through a modal dialog box. Bandwidth configuration, and user operations are recorded. If the user needs to increase the PVC bandwidth configuration, a network interface PVC traffic configuration template for the user to select the required configuration is provided.

Applying this baseline analysis method can effectively warn of poor performance or busy network resources, so as to control the optimization of network configuration. If the user configures a policy server, the network optimization control strategy for certain key network performance indicators (such as the network interface traffic mentioned above) can be customized based on the policy server, and the key network performance indicators are limited to the SLA signed by the user. , so as to realize automatic dynamic optimization control of the network.

The output data using this baseline analysis method can be used as reference data for operators to improve network usage efficiency. For example: for the data in the example, the 24-hour distribution points, use the deviation of 4 continuous distribution points as the judgment mark of serious trend changes. It can be seen that from 0 o'clock to 4 o'clock, the port traffic is low, lower than the baseline value, so the operator can provide some preferential policies for this period, so that users can get a certain discount during this period to make full use of the network.

The network management system using this baseline analysis method and the telecom OSS system establish an interface to transmit some utilization rate performance indicators, which can be divided into different tariff segments for the low utilization rate periods analyzed by the baseline analysis to form a scientific tariff adjustment strategy.

The periodic baseline analysis method can be used to monitor abnormal traffic in the telecommunication network and locate abnormal ports, which is especially important in controlling viruses and black Internet cafes. For example: the network management system that adopts baseline analysis will establish an interface with virus monitoring software, and the ports of abnormal traffic can be found out through traffic baseline analysis, and then further analysis of its transmission packets can find out whether there are packets of DOS attack requests or other virus forwarding servers , so as to carry out port isolation control to prevent it from endangering the operation of the entire network.

Claims (9)

1, a kind of method of monitor network service feature, but described network has the network management system of collection network business datum, it is characterized in that comprising step:

By network management system configuration baseline analysis parameter, if simple baseline, then definite baseline type and historical data are counted, if the periodicity baseline is then determined period type, historical data periodicity and distributed points as analysed basis our unit;

According to disposing the business datum of from network, sampling;

If simple baseline, then from described business datum, obtain mean value, maximum and/or the minimum value that historical data is counted, and be one or more horizontal base line that datum mark forms the reflection service feature with this value, if periodicity baseline, then according to historical data periodicity mean value, maximum and/or minimum value of each distributed points as analysed basis our unit in the acquisition cycle from described business datum, and with this value is one or more baseline that datum mark forms the reflection service feature;

With described baseline is reference, from network, obtain the current business data value and the value corresponding with current time point on the baseline compares, when the business datum value that continuous a plurality of sampled points occur all was greater than or less than the value of corresponding sampled point on the baseline, then there was fixing variation tendency in the service feature of decision network.

2, the method for claim 1 is characterized in that, when there is fixing variation tendency in the service feature of decision network, produces warning information or/and carry out pre-configured control strategy network is carried out optimal control in dynamic.

3, the method for claim 1 is characterized in that, before the maximum and/or minimum value of determining business datum, rejects the abnormal data bigger with the mean deviation of all business datums in the business datum.

4, method as claimed in claim 3 is characterized in that, adopts [mean value-mean value * n%, mean value+mean value * n%] confidential interval come the rejecting abnormalities data, promptly be not regarded as abnormal data in this interval business datum and reject, wherein n% is a confidence level, n=20～30.

5, method as claimed in claim 3 is characterized in that, adopts the mean value of method of weighting computing service data, gives weights near sampled point of current time and is higher than weights from current time sampled point far away.

6, as the arbitrary described method of claim 1 to 5, it is characterized in that, described baseline is for existing the simple baseline of relatively stable characteristic in the expression traffic data is during a time period, this baseline analysis parameter comprises baseline type, historical data point data, the superseded mode of abnormal data and confidential interval.

7, method as claimed in claim 6 is characterized in that, is that datum mark extends with mean value, maximum and/or the minimum value that obtains, and forms the simple baseline of performance perpendicular to the axis of representing the business datum value.

8, as the arbitrary described method of claim 1 to 5, it is characterized in that, there is the period of performance baseline of cyclic variation trend in described baseline within a certain period of time for representing business datum, and this baseline analysis parameter comprises cycle, historical data periodicity, distributional analysis point base unit, the superseded mode of abnormal data and confidential interval.

9, a kind of Network performance monitoring system is characterized in that comprising:

The performance data library module is used to store History Performance Data;

The base-line data library module is used to store baseline configuration data and base-line data;

The service data acquisition module is used for collecting from communication network the real data of relevant performance operational indicator, and stores described performance data library module into;

The baseline analysis configuration module, be used to receive the baseline analysis configuration parameter of customization, if simple baseline, then definite baseline type and historical data are counted, store configuration data into the base-line data library module, if the periodicity baseline is then determined period type, historical data periodicity and distributed points as analysed basis our unit store configuration data into the base-line data library module;

The base-line data analysis module, read History Performance Data and baseline analysis configuration parameter formation baseline from described performance data library module and base-line data library module respectively, if simple baseline, the mean value of counting then according to the historical data that obtains, maximum and/or minimum value, form one or more horizontal base line of reflection service feature for datum mark, and be saved in the baseline database, if periodicity baseline, then according to the mean value of each distributed points as analysed basis our unit in the cycle that obtains, maximum and/or minimum value connect one or more baseline that the back forms the reflection service feature as datum mark, and are saved in the baseline database;

Network optimization analysis module is used for judging the network performance quality according to the business configuration of base-line data, the network equipment and the current performance data that collects, and network is carried out corresponding Optimizing operation.

Images