KR20040028404A - Link Capacity Dimensioning Method for the IP Network with Bursty Traffic - Google Patents

Abstract

PURPOSE: A method for designing a link capacity of an IP network considering bursty traffic characteristics is provided to determine a bursty factor given to a link on the basis of analyzed statistical characteristics, and to determine a design capacity of the link by using the bursty factor, thereby easily designing a future link capacity. CONSTITUTION: A system extracts traffic data(S100). The system calculates average, standard deviation, and intermediate values of the traffic data(S200). The system differently determines a maximum value of a reliable section according to a distribution of statistical data(S300). If the distribution of the statistical data is near to a normal distribution, the system determines the maximum value of the reliable section in a standard normal distribution(S310). The system determines a value obtained by dividing the maximum value with an average as a bursty factor(S400). If the distribution of the statistical data is not near to the normal distribution, the system determines a reliable section by a non-parametric procedure(S320). A bursty factor is a value obtained by dividing a maximum value of the reliable section with the intermediate value of the traffic data(S500). The system multiplies the bursty factor by a predicted value of data, and determines a design capacity of a link(S600).

Description

Link Capacity Dimensioning Method for the IP Network with Bursty Traffic}

The present invention relates to an IP network design technology. In particular, the present invention requires a certain loss rate (α) for traffic imposed on an IP network configuration link providing a best-effort service and is required in the future when providing a service. A method of designing link capacity that is predicted to be.

In addition, the present invention relates to a method for detecting a link causing a bottleneck based on past traffic data and calculating an appropriate capacity required for the link to satisfy a predetermined Quality of Service (QoS) level.

In order to design a link constituting a communication network, it is necessary to know the amount of traffic charged to the link at the time of the design goal. The traffic charged to a specific link at the target time point is generally used as the average value of the estimated traffic estimate. The link design based on the average value does not guarantee proper quality for bursty data traffic.

This is because, even if the average link traffic in the IP network does not exceed the maximum bandwidth of the link, if the fluctuation of the traffic is large, the loss rate exceeds the target level and thus cannot satisfy a predetermined QoS level.

Therefore, it is common to design links in a way that reflects some degree of redundancy.

Conventionally, various studies have been conducted to secure the stability of the network in consideration of such characteristics, but a specific method of calculating the link capacity has not yet been proposed.

In order to solve this problem, the present invention proposes a method for analyzing the statistical characteristics of the link-imposed traffic for calculating the link capacity to satisfy a predetermined packet loss rate for the traffic processed in the IP network providing the best service.

1 is a block diagram of an IP network used for the link capacity design according to the present invention.

2 is a flowchart illustrating a method for designing an IP network link capacity according to the present invention.

Figure 3 is a statistical data graph used in the embodiment according to the present invention.

4 is a flowchart illustrating a method of verifying a quality of service of an IP link according to the present invention.

5 is a flowchart illustrating a process of analyzing statistical data in a method for verifying quality of service of an IP link according to the present invention.

6 is a flowchart illustrating a process of verifying a quality of service of a link and a process of determining a required capacity of the link in the service quality verification method of an IP link according to the present invention.

The present invention relates to a link capacity design method for an IP network including a plurality of links, the first step of analyzing statistical characteristics of statistical data of traffic processed in the link, the link based on the analyzed statistical characteristics And a second step of determining a bursty element imposed on, and a third step of determining a design capacity imposed on the link using the bursty element.

The present invention relates to a link capacity design method of an IP network including a plurality of links. The link operation rate and traffic data, which is a value obtained by dividing the effective link capacity of a link designated by a network operator in a predetermined link by the physical maximum capacity of the link, are provided. , And a first step of analyzing statistical characteristics of the traffic data by using α (packet loss rate of the traffic) as input data, a second step of verifying whether the link is congested, and a link verified to be congested in the second step A third step of determining the appropriate dose of.

Hereinafter, with reference to the accompanying drawings will be described in detail an embodiment of the present invention.

1 shows a general IP network to which the present invention is applied.

The subscriber is connected to the subscriber receiving router 12 through the connection line 11. In the IP network, there are routers having various capacities, such as a backbone network router 13 and a service PoP router 14.

These routers are connected to each other via links 15, 16 and 17 and the present invention relates to a method of designing the capacity of these links and a method of verifying the quality of service of the links.

2 shows a link capacity design method of an IP network according to the present invention.

In this method, using the data extracted in the past, a confidence interval in which congestion occurs below a predetermined packet loss rate (hereinafter, referred to as α) is set, and a burst factor that is numerically quantified by a predetermined statistical method is determined. The link capacity is designed by multiplying the estimated value of the amount of data to be charged to the link.

Each step is described in detail as follows.

First, traffic data for a predetermined period of time is extracted (S100). In this embodiment, the daily busy hour data amount charged to the link in recent months is taken as input data.

The busiest hourly data volume is the largest value among the upward and downward data volumes surveyed by the time of day in the link. Traffic data used in this embodiment is shown in Table 1.

TABLE 1

date Busy hour Data amount (Kbytes) lift Downward Maximum value 20020501 21 1.91361E + 11 6.95834E + 11 6.95834E + 11 20020502 21 1.95303E + 11 7.20316E + 11 7.20316E + 11 20020503 21 1.82244E + 11 7.92388E + 11 7.92388E + 11 20020504 15 1.75725E + 11 7.88808E + 11 7.88808E + 11 20020505 20 1.77088E + 11 7.17853E + 11 7.17853E + 11 20020506 22 1.97159E + 11 7.65117E + 11 7.65117E + 11 20020507 22 2.03855E + 11 7.43338E + 11 7.43338E + 11 20020508 22 2.05339E + 11 7.60705E + 11 7.60705E + 11 20020509 21 1.86646E + 11 7.5558E + 11 7.5558E + 11 20020510 22 1.97255E + 11 7.73797E + 11 7.73797E + 11 20020511 21 1.94579E + 11 8.00409E + 11 8.00409E + 11 20020512 21 1.8802E + 11 7.4078E + 11 7.4078E + 11 20020513 00 1.57386E + 11 5.5397E + 11 5.5397E + 11 20020514 21 1.92494E + 11 6.56121E + 11 6.56121E + 11 20020515 14 1.97827E + 11 9.04347E + 11 9.04347E + 11 20020516 22 1.98918E + 11 7.87463E + 11 7.87463E + 11 20020517 22 1.96428E + 11 7.93882E + 11 7.93882E + 11 20020518 20 2.02364E + 11 8.49145E + 11 8.49145E + 11 20020519 20 2.08414E + 11 7.62862E + 11 7.62862E + 11 20020520 00 1.69764E + 11 5.7898E + 11 5.7898E + 11 20020521 21 1.91758E + 11 8.09584E + 11 8.09584E + 11 20020522 21 1.89871E + 11 7.08946E + 11 7.08946E + 11 20020523 21 1.92614E + 11 7.38847E + 11 7.38847E + 11 20020524 21 1.8872E + 11 7.39132E + 11 7.39132E + 11 20020525 21 1.77174E + 11 7.59936E + 11 7.59936E + 11 20020526 22 1.73847E + 11 5.1538E + 11 5.1538E + 11 20020527 22 1.67742E + 11 5.12798E + 11 5.12798E + 11 20020528 21 1.73245E + 11 5.20419E + 11 5.20419E + 11 20020529 21 1.68093E + 11 6.28599E + 11 6.28599E + 11 20020530 22 1.73616E + 11 7.78221E + 11 7.78221E + 11 20020531 23 1.73912E + 11 7.4135E + 11 7.4135E + 11

Such data can be measured at each router using a simple network management protocol (SNMP). The traffic data of a specific day represents the data amount of the up-peak time and the down-peak time respectively, and the up-peak time and the down-peak time may be different. The maximum value represents a larger value between the uptime data amount and the downtime data amount.

In the analyzing of the traffic data (S200), the average, standard deviation, and median of the daily busy hour data amount for one month are calculated with reference to the traffic data.

In the next step, the upper limit of the confidence interval is obtained from the daily data amounts and the burst factor is determined using the upper limit value.

Since the packet loss rate is α, the confidence interval is an area with a probability of (100-α). The upper limit of the confidence interval is determined differently according to the distribution of the statistical data (S300).

If the distribution of the statistical data is close to the normal distribution (S310), the upper limit of the confidence interval is determined using z _{α / 2} satisfying P (Z> z _{α / 2} ) = α / 2 in the standard normal distribution.

The value converted from the actual normal distribution is the upper limit of the confidence interval, and the value becomes {mean + z _{α / 2} × (standard deviation)}. The value obtained by dividing the upper limit of the confidence interval by the average is determined as the bursty factor (S400).

If the distribution of the statistical data is not close to the normal distribution, the (100-α) confidence interval is determined by a nonparametric procedure (S320).

In this case, the bursty element is a value obtained by dividing the upper limit of the confidence interval by the median of the data (S500).

For reference, the nonparametric method is one of the statistical methods of hypothesis testing without making any assumptions about the parameters (mostly about the distribution of the population).

In nonparametric methods, the statistics used for hypothesis testing are rank statistics, and the most commonly used statistics are median. The hypothesis testing of the median of the population and the confidence intervals can be obtained. The most commonly used method is the Sign Test (or Ranked sign test).

The nonparametric method is a commonly used statistical method, and a detailed description thereof will be omitted.

3 is a graph showing the distribution of data amounts shown in Table 1. FIG. This data does not exactly match the normal distribution, but it shows that the distribution is close to the normal distribution. The results of obtaining the bursty elements by the above-described method are shown in Table 2 below.

TABLE 2

Average Standard Deviation Upper confidence interval Bursty element 7.22416E + 11 97812310811 1.02466E + 12 1.42

In the third step, the design capacity of the link is determined using bursty elements. In the present embodiment, the design capacity of the link is determined by multiplying the bursty factor with an estimate of the amount of data to be applied to the link (S600).

The predicted amount of data to be charged to the link means the amount of data set by predicting future service demand.

4 is a flowchart illustrating a service quality verification method of an IP link according to the present invention.

In the first step, the traffic data measured for a certain period, the packet loss rate, and the link operation rate determined by the network operator are input, and the traffic data is analyzed (S10). The link operation rate is a value obtained by dividing the effective link capacity designated by the network operator by the physical maximum capacity of the link.

In the second step, the amount of data processed by the link is processed using the analysis result in a statistical manner to verify whether each link satisfies the service quality level (S20).

In the third step, for the link that does not satisfy the service quality level, the link calculates an appropriate amount of data for simultaneously satisfying a predetermined service quality level and a link operation rate (S30).

5 shows a process of analyzing statistical characteristics of traffic data.

In this process, the effective link capacity E is calculated based on the link operation rate determined by the network operator (S11). Effective link capacity is determined by E = (maximum link capacity) × (link utilization). This value is the actual maximum capacity for the individual link.

As described above, the link operation rate is a ratio of the actual effective link capacity determined by the network operator within the maximum capacity (bandwidth) of the link. If the link operation rate is 100%, the physical link capacity is used.

Statistical characteristic analysis in this embodiment is a process of analyzing the mean and standard deviation of the statistical data (S12, S13). If n is the traffic measurement days and x _i is the daily busy hour traffic, the average (B) and standard deviation (D) are expressed by the following equation.

[Equation 1]

[Equation 2]

6 is a flowchart illustrating a process of verifying a quality of service of a link and a process of determining a required capacity of the link in the service quality verification method of an IP link according to the present invention.

In step S21, the limit coefficient X is calculated using a Gaussian approximation for a section within a packet loss rate that is a service quality target level. The upper limit of the confidence interval of X = {(1-packet loss rate)-B} / D is determined.

Step S22 is a step of verifying the congestion probability of the link, and determines whether the effective capacity E of the link satisfies the conditional expression E> B + XD. If the condition is met, the link is a link that satisfies the quality of service target.

If the conditional expression is not satisfied, the link is a bottleneck link that does not satisfy the QoS.

For the bottleneck link, the necessary link capacity Y for calculating the service quality level is calculated (S31).

The required link capacity is given by Y = (B + XD) / (link utilization), which is the minimum bandwidth to meet a given quality of service level.

According to the present invention, it is possible to easily design future link capacity when determining a predetermined expected packet loss rate for a specific link and an estimate of the amount of data to be processed in the link in the future.

In addition, according to the present invention, it is possible to verify whether or not a runaway link is congested based on a predetermined quality of service, and the link capacity required to satisfy the quality of service can be easily calculated for the runaway link.

The embodiments of the present invention are for the purpose of illustration and those skilled in the art can make various modifications and changes of the embodiments through the technical spirit of the claims, and such modifications and changes are within the scope of the claims.

Claims (9)

In the link capacity design method of an IP network including a plurality of links, Analyzing a statistical characteristic on statistical data of traffic processed in the link; Determining a bursty element imposed on the link based on the analyzed statistical characteristic; And A third step of determining the design capacity imposed on the link using the bursty element Link capacity design method of an IP network comprising a. The method of claim 1, The statistical data of the traffic is the daily busy hour data amount imposed on the link for a predetermined period of time based on the time point of performing the analysis, and the daily busy hour data amount is the most of the upstream and downstream hourly data amounts on the link. A method for designing link capacity of an IP network, characterized by representing a large value. The method of claim 1, In the first step, the process of analyzing the statistical characteristics And determining an average, standard deviation, and median value of the traffic data. The method of claim 1, wherein the second step Examining whether the statistical data can be approximated with a normal distribution, When the statistical data can be approximated with a normal distribution, {(1-α) confidence interval upper limit value} / (average value of the statistical data) is determined as a burst factor, and the statistical data can be approximated with a normal distribution. If not, determining {(1-α) confidence interval upper limit value} / (middle value of the statistical data) as a bursty element Wherein α is a packet loss rate of the traffic. The method of claim 1, In the third step, the design capacity is determined by multiplying the bursty factor by an estimate of the average amount of data of the link. In the link capacity design method of an IP network including a plurality of links, Statistical characteristics of the traffic data by inputting link operation rate, traffic data, and α (traffic packet loss rate), which is a value obtained by dividing the effective link capacity of a link designated by a network operator on a predetermined link by the physical maximum capacity of the link. Analyzing the first step; A second step of verifying whether the link is congested; And A third step of determining an appropriate capacity of the link verified to be congested in the second step Quality of service verification method of the IP network comprising a. The method of claim 6, Statistical analysis of the traffic data of the first step is a process of determining the average and the standard deviation of the traffic data, the traffic data is the daily busy amount of data charged to the link for a predetermined period based on the analysis time And the daily busy time data amount is the largest value among uplink and downlink data amounts for each time slot in the link. The method of claim 6, wherein the second step If the effective link capacity designated by the network operator for the link is larger than {1 + (limit coefficient) x (standard deviation of the statistical data)} of the statistical data, it is determined that the link is not congested. The coefficient is determined by the upper limit of the confidence interval of (limit coefficient) = {(1-α) minus (average value of the statistical data)} / (standard deviation of the statistical data). . 7. The method of claim 6, wherein the appropriate link capacity of the third step is {(The mean value) + (the limit coefficient) x (the standard deviation)} of the statistical data is determined as a value obtained by dividing by the link operation rate, wherein the limit coefficient (limit coefficient) = confidence of {(1-α) The upper limit value of the interval-(average value of the statistical data)} / (standard deviation of the statistical data) characterized in that the service quality verification method of the IP network.