CH651681A5 - METHOD FOR OPERATING A DATA PROCESSING SYSTEM WITH A COMPUTER. - Google Patents

Description

The present invention relates to a method for operating a data processing system, the features of which are specified in the head of claim 1.

A data processing system is known, see e.g. Swiss Patent No. 613 539. This system is particularly suitable for overload regulation or control in a telecommunication switching network. In the known system, the workload control means compare the determined degree of occupancy of the computer with a minimum and a maximum value at the end of each control period and reduce or increase the maximum acceptable workload, i.e. for example the acceptable number of new calls that can be accepted for processing during such a regulation period, or they receive this workload constantly depending on the determined occupancy rate, which may be above the maximum level, below the minimum level or between the two levels . The various acceptable workload values form a set of predetermined discrete values, the changes of which are consequently made in stages.

A disadvantage of this known system is that under certain circumstances the processing of a number of new calls (ie new load) could be unnecessarily delayed, although the desired maximum occupancy of the computer has not yet been reached, which is meant by this maximum degree or level for which all operating criteria are still met. Just assume that the number of calls offered for processing after the end of a regular period, in which the degree of occupancy of the computer and the actually accepted number of calls are well below the maximum occupancy and the acceptable number of new calls to be processed during this regular period lay, increases considerably. Such an increase in the number of new calls offered can e.g. can be attributed to a defective inter-office multi-conductor cable, which causes a high number of line loop closures, which simulate a corresponding number of incoming calls. In this case, the number of calls accepted by the computer for processing will increase considerably during the next control period, which very quickly leads to the maximum permissible value being reached for these periods; however, the computer occupancy level is generally not increased. In fact, after discovering such new incoming calls and accepting them for processing, the computer will soon cease the processing operation because for these calls no preparatory character was received from the remote office, which character would cause further processing, so that only one for each wrong call is used for a short time and consequently a very large number of these calls can be handled. This means that a large number of incorrect new calls are processed, which leads to a delay in the processing of a number of normal calls. Since the occupancy level remains constant or even increases slightly, but remains below the maximum level, as was assumed, the permissible maximum number of new calls to be accepted is unchanged.

The purpose of the present invention is to avoid the disadvantages mentioned, reference being made to the features according to the invention which are specified in the characterizing part of patent claim 1.

Within certain limits, the acceptable workload depends not only on the determined degree of occupancy of the computer, as in the known systems, but also on the amount of workload currently being processed by this computer. For example, in the case mentioned above, in which the number of calls processed increases while the occupancy level remains practically the same, the calculated acceptable workload will be greater than that calculated for the previous period. This means,

that more workload can be accepted for processing.

Other congestion control systems for telephone traffic, e.g. in U.S. Patents 3,517,123 and 3,623,007 as well as in the article "Real-time testing of automatic overload control systems in a laboratory environment" by F.T. Mann, published in IEEE Transactions on Communications in September 1978, pages 1027-1031, deal in particular with a programmed electronic traffic simulator and describe strategies by means of which the workload of a computer is modified in accordance with its degree of occupancy. The workload is limited by limiting the number of new calls to be processed during a fixed time interval. In US Pat. No. 3,517,123, this is accomplished by excluding the computer from responding to all requests for service from subscribers, except for preferred services. As with the aforementioned Swiss patent, however, the number of new calls accepted for processing during the immediately preceding fixed time interval is not taken into account either.

Even before the introduction of dynamic congestion controls for local traffic in telephone exchanges, this control was used to redirect calls in order to avoid blocking offices - see e.g. U.S. Patent 3,796,837. However, these systems do not depend very much on the use of computerized offices.

Also, the known redirection techniques do not require the calculation of new workload limits that would be based both on actual values received earlier and on the current degree of occupancy.

An embodiment of the invention will now be explained in more detail with reference to the drawing, which represents a schematic representation of a data processing system.

This data processing system forms part of one

5

10th

15

20th

25th

30th

35

40

45

50

55

60

65

651 681

automatic telecommunications switching network, which additionally contains an automatic switching network TSE. The data processing system has a computer with a memory M EM and a processor PR.

The computer can collect call information from the automatic switching network TSE and accordingly control the establishment, monitoring and triggering of call connections by the switching network by successively carrying out higher and lower priority tasks or clock or basic programs. The computer begins executing this higher priority task in every so-called clock interrupt interval of 20 ms; if these tasks are completed and there is still time available, the computer begins executing lower priority tasks. Typically, the lower priority tasks are completed before the start of the next 20 ms clock interval. As a result, the computer has a certain amount of free time per cycle interruption interval of 20 ms and with a normal number of tasks performed. When the number of tasks performed by the computer is increased, the free time naturally becomes shorter. This free time is therefore a measure of the tasks performed by the computer and therefore also a measure of its occupancy level or degree.

Some of the tasks performed by the computer are dedicated to the processing of calls, while another part - the so-called "fixed" or "general part" - is independent of the call processing, these fixed-part tasks being carried out even with no call traffic at all will. Since the time required to carry out the “fixed part” functions is fixed, only changes in the call processing time can lead to changes in the computer leisure time and consequently the degree of computer occupancy. In order to be aware of these changes, the degree of occupancy of the computer is determined; This expresses the number of new calls that can be accepted by the computer during a specific, fixed time interval, or more generally; new workload, calculated and then limited to the calculated number mentioned. The process will now be described further.

The computer contains means for periodically determining its degree of occupancy during a fixed period of a duration T of e.g. 4 seconds. This is done by measuring the total free computing time Fj at the end of each such regulation period P, -, which was available during this period P, - and by calculating the degree of occupancy Oj for this regulation period using the following equation (1).

The computer further contains control means which, after the determination of the degree of occupancy Oi by the determining means likewise with a periodicity T, a provisional value Npj + 1 of the acceptable number of new calls for processing during each of the control periods of the period P; immediately following control period Pi +, calculate; The tax periods are fractions of the regular period and have a duration of e.g. T / n = 1 second (at n = 4).

The expression Npj + i is calculated from the following equation:

° M "V

np, I + i = tty (2)

where Np.j + i and Oj are defined above,

V is the «fixed part» of the work performed by the computer,

M; indicates the average number of new calls that were actually accepted for processing during each of the four control periods of each control period Pi, and 0M indicates the maximum permitted value of the degree of occupancy of the computer.

In the above-mentioned equation (2), the value V of Om and O; subtracted because the number of calls to be processed or processed is independent of V. Om-V is a value corresponding to the maximum time available for such processing, and Oi-V is a value corresponding to the time used for such processing during the control period Pj.

The provisional value Np> j +) thus calculated, after being rounded down to the next whole number, is used as the new value Ni +) for the control period Pj + i, if Npj +), Oj and M; with respect to the limit values (1 - b) N ;, (1 + a) Ni, Nj + 1; Om, Oe; fNj have predetermined values.

Here means:

a a fraction such that aN; the maximum allowable

Increase of Nj + 1 with respect to N; indicates;

b a fraction such that bNj is generally the maximum allowable reduction of N; +1 with respect to Nj;

f a fraction such that fN; is a minimum value of the number of calls acceptable during a tax period;

Om the minimum value of the occupancy of the computer; Oe the excessive occupancy rate (see further description).

The rules for choosing the final new value Nj + | for the control period Pj + i are as follows:

If Oj <Om and additionally M; <fNi, the new value Ni + 1 becomes equal to the value N; regardless of the calculated value Npj + | be; if either Oi <Om or Mj> fN; and in addition O; <Oe, then N, +1 is (1 - b) Nj,

where (1 + a) Ni or Np.i + | depends on whether Np, i + | less than (1 - b) N; or is greater than (1 + a) Ni or is between these limits. However, if Np_i + 1> Ni + (and aN | <1, then Ni + | is N; + 1; if Oj> Oe, then N, - +) is (1 - b —d) Nj, or (1 - d) Np.; +), depending on whether NPii +1 is less than (1 - b) Nj or between sizes (1 - b) N and (1 + a) Nj. Here d is a fraction that allows the calculation of the additional reduction dNi or dNp i +1 and b is a fraction that allows the calculation of a normal reduction. If Oi> Oe, then Np j +1 is always smaller than Nj (1 + a), since, according to equation (2) and according to the expression Oe> 0M, the former expression is already smaller than Mj, which has the value Nj cannot exceed.

The above-mentioned rules are summarized in the table below, in which yl to y7 are the results of the following inequalities, in which y2-y3 and y6 * y7 represent the logical AND combination of y2 and y3 or y5 and y6 and in which yl to y7 are set to 1 if the corresponding inequality is correct and set to 0 if it is wrong. The expression x means that the value from yl to y 7 has no influence on the choice of Nj + i.

(1 + a) Nj

<

Np.i + j yi

(3)

Ni + 1

<

Np.j + 1

1

y2

(4)

aNj

<

y3

(5)

N p.i + I

<

(1 -b) Ni y4

(6)

O.

<

Oj y5

(7)

Oj

<

om y6

(8th)

M;

<

fNj yi

(9)

4th

5

10th

15

20th

25th

30th

35

40

45

50

55

60

65

5

651 681

yi y2-y3

y4

y5

y6-y7

Nj + I

X

X

X

X

1

Ni

X

X

1

0

0

(l-b) Ni

0

X

0

0

0

Np, i + I

1

0

X

0

0

. (l + a) Nj

X

1

X

0

0

Ni + i

X

X

1

1

X

(l-b-d) Ni

0

X

0

1

X

(1 - d) Npj + i

The parameters mentioned above can have the following values, for example:

0m = 0.75 a = 0.1

Om = 0.95 b = 0.015

Oe = 0.975 d = 0.1

V = 0.40 f = 0.75 T = 4 s T / n = 1 s

With regard to the parameters, the following should also be noted:

The duration T / n of the tax period

On the one hand, this duration should not be too short, since in this case only a correspondingly small number of new calls can be accepted, which could lead to increased delays in call processing if the number of calls offered is high. If the tax period is extremely short and the acceptable number of calls results in the occupancy rate of the computer being considerably lower than its maximum occupancy level 0M, an increase of 1 in the latter number can result in an occupancy rate above 0M.

On the other hand, this duration should not be too long, since in this case a correspondingly large number of new calls could be accepted; this would occur in the first 20 msec interruption intervals of the control period and would cause a significant spike in the occupancy of the computer. In order to defuse this state, the number of calls accepted per 20 ms cycle interruption interval can be limited, e.g. in the already mentioned article by F.T. Man has been described.

The duration T of the regular period

On the one hand, the duration should not be too short, since a random decrease in the number of calls Mj actually accepted for processing would result in a corresponding decrease in the maximum accepted number of calls Ni, because the degree of occupancy in the computer does not immediately reflect this decrease by lower values . In this case the processing of the calls is delayed.

On the other hand, this duration should not be too long since the computer would not react promptly to abrupt changes in the number of calls offered for processing.

A too rapid increase in Mi due to random spikes in the number of calls offered for processing does not cause any problems since Mj is limited by Nj.

The maximum occupancy level is 0M

It is the maximum desired or permitted occupancy of the computer, i.e. a degree of occupancy at which a maximum number of calls can be processed during a fixed time interval, the computer still meeting all other operating criteria. 0M becomes e.g. chosen by 3% lower than the maximum possible, i.e. maximum permitted, occupancy rate. Setting 0M too low results in degraded service levels during periods of high traffic because calls that could have been satisfactorily processed by the computer are uselessly delayed.

The excessive occupancy rate Oe

This is a dangerously high value for the occupancy rate, which should only be exceeded during very short time intervals. The value is just below the maximum possible, i.e. maximum permitted, occupancy rate.

The minimum occupancy Om

The value of this parameter is e.g. 0.4.

The «fixed part» V

This value depends on the type of switching matrix and is e.g. 0.4.

The parameters a + b

These parameters are used to prevent excessive fluctuations in Nj, e.g. due to a different size of incoming calls or through the call type distribution, i.e. can be caused by the call mix. In practice, a is considerably larger than b, since while it is useful to strictly limit the value of the increase in Nj, in most cases a better level of work is achieved by allowing a more open increase in Ni. The reason for this asymmetry follows from what has been mentioned in connection with the duration of the control period.

The parameter d

As explained above, this parameter is used to calculate an additional reduction in Ni.

The parameter f

The value of this parameter is e.g. 0.75. The data processing system is explained in more detail below with reference to the drawing.

The memory M EM of the computer contains a number of memory cells SL 1 to SL 28, while the processor PR has a bistable circuit BS, 4 clock generators CL 1-CL 4, a comparator CR, two registers RE 1 and RE 2 and a call processing and Has control arrangement CPA. This arrangement CPA is via a connection c 1 to the switching network TSE, via a connection c 2 to the memory MEM, via connections c 3 or c 4 to the register RE 1 or RE 2, via a connection y to the comparator CR, via connections c 5 -c8 connected to the clock generators CR 1-CR 4 and finally via a connection c9 to the 1 input of the bistable circuit BS. The function of CPA is clearly evident from the description of how the system works.

The content of the memory MEM is taken into account, e.g. at the beginning of the last or 50th cycle interruption interval of 20 ms for the last (CP 4) of the 4 control periods CP 1-CP 4 of the i-th control period Pi.

The following data are then stored in the memory MEM:

- The above-mentioned constant parameters Oe, 0M, a, b, d, f, V and also 1. These parameters were stored in the memory by the arrangement CPA via connection c2 at the beginning of the operation of the plant;

a number of variable parameters Nj, Ni + |, (1 + a) Nj, (I -b) Nj, aNj, fNj, Oj_i, Mj_t, NPii and yl to y7. These parameters were stored in the memory by the arrangement CPA via connection c2 at the end of the previous control period Pj_i and have the following meaning:

5

10th

15

20th

25th

30th

35

40

45

50

55

60

65

651 681

6

- N [is the maximum still acceptable number of new calls to be processed for each of the 4 successive control periods CP1 to CP 4, which is the control period P; form;

- Oj_i is the occupancy level of the computer during the regular period P; _! ;

- Mj_ i is the average number of new calls that were actually received for processing for each of the 4 successive control periods CP1 to CP 4 of the control period P, -_ f;

- NPj is the calculated, provisional number of new calls that are to be accepted for processing during the period P ,.

The following further parameters are also stored in the memory MEM:

- M, i.e. the number of new calls which have already been accepted for processing during the current control period CP 4 of the current control period Pi;

- S, i.e. the total number of new calls that have already been accepted for processing during the current control period Pi;

- F, i.e. the free time that was already counted during the current regular period Pj.

The comparator CR compares the values stored in the registers RE 1 and RE 2 and switches its output y on (y = 1) if the value stored in RE 1 is smaller than that in RE 2.

The clock generator CL 1 generates output pulses at intervals of 20 ms and thus defines the 20 ms clock intervals in the arrangement CPA via the connection c5. The clock CL 1 synchronizes the clock CL 2 via the connection clO and the bistable circuit BS is reset via the connection cl 1.

The clock generator CL 2 generates output pulses in 1-second intervals and thus defines the 1-second control periods in CPA via the connection c6. The clock generator CL 3 synchronizes the clock generator CL 2 via the connection cl2.

The clock generator CL 3 generates output pulses in 5-second intervals and thus defines the 5-second control periods in CPA via the connection c7.

Finally, the clock generator CL 4 is connected to the 1 output of the bistable current circuit BS via the connection cl3 and generates output pulses at intervals of 25 microseconds when SPA has triggered the bistable circuit BS into its 1 state via the connection c9. These output pulses are applied to CPA via connection c8.

Each of the connections cl to cl3 has one or more wires, depending on the signaling requirement.

In order to make the explanations clearer, it is first described how the degree of occupancy of the computer at the beginning of the control period Pj +) is determined by the determination means which contain the clock generators CL 1, CL 3, CL 4 and the bistable circuit BS and which by the call processing and control arrangement CPA cooperating with the memory MEM.

At the beginning of each 20 ms clock interruption interval of this control period Pj, the arrangement CPA begins with the execution of a series of tasks. After completion of these tasks and before the end of each of these intervals, the arrangement CPA sets the circuit BS into its 1 state via the connection c9. As a result, the circuit BS excites the clock CL 4, which generates output pulses with a periodicity of 25 microseconds for counting purposes. Each time the connection c8 is switched on by such a pulse, the arrangement CPA responds to this process by storing the memory location SL 28 of the memory

MEM addressed via the connection c2 in order to obtain the free time F stored there. This free time is then increased by 1 and stored again in the same memory cell SL 28. At the beginning of the next 20 ms cycle interruption interval, the circuit BS is reset to its O state by the clock generator CL 1 via the connection C11. The memory location SL 28 thus collects the free time F that has elapsed since the last task of a series of tasks was carried out for each 20 ms cycle interruption interval. This free time F is stored in units of 25 microseconds.

At the beginning of every 20 ms cycle interruption interval, which is defined by the clock CL 1, the one-second clock CL 2 is synchronized and after the elapse of a one-second control period, the clock CL 2 switches on its outputs c6 and c! 2. Each time connection c6 is switched on at the end of such a one-second control period, the arrangement CPA responds by deleting the content of the memory cell SL 26, since the character M stored therein always indicates the number of new calls which have already been accepted for processing during the current control period.

At the beginning of each one-second control period defined by CL 2, the 5-second clock CL 3 is synchronized; after the 5-second control period has elapsed, e.g. Pj, CL 3 switches on its output cl every time. The arrangement CPA reacts to this activation of c7 by reading out the free time F, which was stored in the memory cell SL 26 of the memory MEM via the connection c2, and calculates the occupancy rate using the previously given equation (1), in which T = 4 seconds is expressed in the same units as F, ie in units of 25 microseconds while F is Fj since the end of Pj has been reached.

In the following, it will now be described how the workload regulating means containing the clock generator CL 3, the registers RE 1, RE 3 and the comparator CR 2, which are controlled by the arrangement CPA cooperating with the memory MEM, the size Ni +1 for calculate the period Pi + 1.

At the beginning of the aforementioned 50th cycle interruption interval of 20 ms for the fourth one-second control period of the i-th control period, the arrangement CPA begins with the execution of a series of higher priority tasks. One of these tasks involves collecting data about new calls from the switching equipment TSE over the connection cl; every time a new call is detected, a check is carried out to determine whether this new call can still be accepted for processing or not.

For this purpose, the arrangement CPA responds to the detection of such a new call by reading out the sizes M and N; from the memory cells SL 26 and SL 10 of the memory MEM via the connection c2, by subsequently increasing M by 1, by applying the increased value of M and the value of Nj to the registers RE I and RE 2 via the connection c3 or c4 and finally by checking the state of the output y (0 or 1) of the comparator CR:

If y is in its 0 state, which indicates that the number of new calls accepted would be greater than Nj, if this new call were accepted, this new call will not be accepted for processing. This also applies to the new calls identified by Pj during the CP 4 tax period. The system can of course be designed so that priority calls are always accepted, see e.g. U.S. Patent 3,517,123.

- When y is in its 1 state, which causes

5

io

15

20th

25th

30th

35

40

45

50

55

60

65

7

651 681

it is indicated that the number of new calls accepted is still greater than Nj, CPA stores the increased value of M in the memory cell SL 26 via the connection c2.

The value of M indicates the number of new calls that have been counted since the start of the control period CP 4, since CPA was informed of this start by CL 2 via the connection c6 and upon receipt of this information by deleting the content of the memory cell SL 26 responded.

Simultaneously with the information about the end of the control period Pj, which is provided by the clock generator CL 3 to the CPA via the connection c7, the arrangement CPA performs the following functions, whereby it should be noted that at this moment the parameters S and F values Sj and Fj achieved:

- It calculates the degree of occupancy Oj in a manner which has already been described above and stores this value in the memory cell SL 16 of the memory MEM by overwriting the previous value 0; _:;

it reads Sj from memory cell SL 27 and calculates Mj = Sj / 4, since Mj is the average number of new calls accepted by Pj for processing during this of the 4 control periods CP 1 to CP 4;

- It reads the values Om and V from the memory cells SL 2 and SL 9 of the memory MEM and uses these values and the values Mj and Oj to calculate the value Npj + b given by the above-mentioned equation (2);

- It stores the calculated values of Mj and Np, i + j in the memory cells SL 17 and SL 18 by overwriting the values M; _i and Np> i previously stored there.

The arrangement CPA then checks whether the calculated provisional value Npj + | can be assumed as the new value Mi + i by calculating the limits of Npj +:, O, and Mj. For this purpose CPA reads the values (1 + a) Nj and Np.j + | from the memory cells SL 12 and SL 18, feeds them into the registers RE 1 and RE 2 and subsequently stores the state yl of the output y of the comparator CR, which compares (1 + a) Nj and Np.i + i - see Equation (3) above - in the memory cell SL 19 by overwriting the value previously stored there.

The arrangement CPA also reads the value 5 pairs Np.i + i, Ni + 1; aNif 1; NpJ + !, (1 -b) N ;; Oe, Oj; Oj, Om; and Mj, fNj and stores the results y2 to y 7 of the corresponding comparisons (4) to (9) in the memory cells SL 20 to SL 25 by overwriting the values previously present there.

The arrangement CPA has consequently calculated the results yl to yl of the above-mentioned expressions (3) to (9) and stored them in the memory MEM. Subsequently, these results are checked, and on the basis of this check, the value of Ni + 1 is selected as shown in the above-mentioned Table 15, where No i + is the previously calculated value.

Finally, CPA calculates the values of Ni + | (l + a) Ni + |, (1 -b) Ni + j, aNi + i, fni + | and stores Ni + | and the mentioned values in the memory cells SL 10 to SL 15.

It is evident from the above considerations that the determination means determine the degree of occupancy Oj of the computer periodically, with a periodicity of T, and that the workload control means, likewise with a periodicity of T, periodically determine a provisional value Np.i +: der calculate the maximum allowable workload processed by the computer. 25 The last-mentioned provisional value Np-i + 1 is directly dependent on and changes in the same way as the desired maximum occupancy 0M of the computer and the registered amount Mf of the workload processed by the computer. This provisional value Np.j + 1 is then assumed to be the new 30 value Nj + 1 of the maximum permissible workload if the provisional value Np,; + 1, Oj and Mj are predetermined values with respect to the limits (1 - b) Nj, (I + a) Nj; Nj + 1, Om, Oe and fNj.

Instead of measuring the degree of occupancy of the computer in the manner described above, one can follow the explanations of US Pat. No. 3,623,007.

G

1 sheet of drawings

Claims (16)

651 681 PATENT CLAIMS (1 -b) Nj <Np.i + l <(1 + a) Ni, where parameter d is positive and less than 1. (1-b) Ni <Np.i + | <(1 + a) Nj and Oi> Om or Mj> fN; applies. 1. A method for operating a data processing system, containing a computer processing a workload, further means for registering the amount of a computer during a certain period (P ^ for processing newly accepted workload Nj, means for periodically determining the degree of occupancy Oj of this computer, wherein the degree of occupancy indicates the effective occupancy of the computer by the respective workload in relation to its maximum permitted occupancy, and finally means, controlled by the aforementioned means of determination, for periodically regulating the acceptable workload Ni + 1 of the computer, the subsequent one for the period (Pj) mentioned Period (Pj + i) is to be accepted again, characterized in that the mentioned control means in the second period (Pj +1) the acceptable workload Ni + | depending on the one hand on the quantity of new ones accepted for processing in the first period (Pj) Workload Nj and, on the other hand, calculate the occupancy Oj determined from the first-mentioned period (P;) within predetermined limits in such a way that the tendency to change is maintained at the acceptable workload Ni + 1 and that of the occupancy Of is reversed. 2nd 2. The method according to claim 1, characterized in that within the said predetermined limits, the acceptable workload Ni + | is directly dependent on the maximum permitted occupancy 0M of the computer and changes in the same sense as this. 3rd 651 681 3. The method according to claim 2, characterized in that the means for regulating the acceptable workload periodically on the basis of each fraction, i.e. calculate the acceptable workload Ni + | for the second period (Pj + i) in each tax period (CP1-CP4), the first mentioned period (PO determined amount of new workload Nj). 4. The method according to claim 3, characterized in that the said calculation of the acceptable workload N; + i is carried out by determining a preliminary value Np i + i of the acceptable load Ni + i, which is dependent on the average during each of the Tax periods of the first-mentioned control period (Pj) for processing the assumed amount of new workload Ni and of the maximum permitted occupancy 0M of the computer and changes in the same way as the mentioned amount of new workload Ni, and that the means for regulating the workload also calculate the predetermined limits and accept the provisional value Np i + as the acceptable value if the mentioned limits are observed. 5 5. The method according to claim 4, characterized in that the provisional value Np.i +1 at the beginning of the second-mentioned control period (Pj + i) is calculated according to the following equation: ° m "v Np, i + 1 = <Mi> ÖT ^ T» in what equation Mj is the average of the amount of new workload actually accepted for processing for each of the Tax periods of the first-mentioned regular period (Pj), Om the maximum permitted occupancy rate, Oj the degree of occupancy determined for the former Control period (Pj) and V is a constant smaller than Oj. 6. The method according to claim 5, characterized in that for each of the control periods of the first-mentioned control period (Pj) its acceptable workload Ni is calculated, and that this workload has limits which are defined by the following expressions: (1 - b) Nj and ( 1 + a) Ni, where a and b are positive and less than 1. 7. The method according to claim 5, characterized in that the degree of occupancy Oj has a lowermost Om and an uppermost Oe limit, Oe being greater than Om. 8. The method according to claim 5, characterized in that the lower limit of the amount of workload Mj actually accepted for processing is defined by the expression fNj, where f is positive and less than 1. 9. The method according to claims 7 and 8, characterized in that the acceptable workload Nj + i of the second control period (Pj + i) is equal to the assumed workload Ni of the first control period (Pj) preceding the second control period, if Oi <Om and Mj <fNj. 10th 10. The method according to claims 6, 7 and 8, characterized in that Nj + i = (1 - b) Nj is if at the same time Np.i + i <(1 - b) Nj and Oj> Om or Mj> fNj. 11. The method according to claims 6, 7 and 8, characterized in that Nj + i = Np.i + i is if at the same time 12. The method according to claims 6, 7 and 8, characterized in that Nj + | = (1 + a) Nj is if at the same time Oj <Oe and aNj> l or Np.i + l <Ni + | and Oi> Om or M;> fNj. 13. The method according to claims 6, 7 and 8, characterized in that Ni + 1 = Nj + 1 is when at the same time Oj <Oe and Np.i + i <Nj + | and aN; <l and Oj> Om or M;> fNj. 14. The method according to claims 6, 7 and 8, characterized in that Ni + i = (1 - d) Np.j + 1 is if at the same time Oj> Oe and 15. The method according to claims 6, 7 and 8, characterized in that Nj + i = (1 - b - d) Nj is if at the same time 0,> 0e and (l-bJNi <Np.i + | - where parameter d is positive and less than 1. 15 20th 25th 30th 35 40 45 50 55 60 65 16. Application of the method according to one of the preceding claims for processing the call traffic in an automatic switching network (TSE), the data processing system controlling the switching network (TSE), characterized in that the said workload includes the processing of the call traffic, and that the data processing system provisional acceptable number of new calls Npj + | processed during the tax period (CP1-CP4) of the second-mentioned control period (Pi + i) from an average number of new calls Mj accepted for processing during the control periods of the first-mentioned control period (Pj) and at the same time a further part-time (V) of each control period of the Call processing dedicated to performing other tasks independently.