1、m 2326434 0005253 5 m Page 1 0 TKS62-a4 Recommendation T/CS 62-04 (Montpellier 1984, modified in Odense 1986) PERFORMANCE AND AVAILABILITY DESIGN OBJECTIVES FOR DIGITAL LOCAL AND COMBINED EXCHANGES Recommendation proposed by Working Group T/WG 1 1 “Switching and Signalling” (CS) “The European Confer
2、ence of Postal and Telecommunications Administrations, considering - that the achievement of long-term harmonisation of integrated digital networks and integrated services digital - the importance of specifying performance requirements appropriate to the integrated digital network and inte- networks
3、 represents a major target in the current development of the European network, grated services digital network, recommends that the members of the CEPT, when introducing digital local and combined exchanges, take account of and give effect to the following performance requirements specifications.” T
4、he reader is referred to CCITT Recommendation 4.514 i, to which this Recommendation is identical, except for the additions and modifications indicated below. The section references (indicated Q) are to 4.514 i. I, Jirst paragraph, last sentence: The reference is to CEPT Recommendation CEPT T/CS 62-0
5、1 2. I, replace last paragraph with the following two paragraphs and note: These design objectives may be used as guides in system design and by Administrations and RPOAs in evaluating specific designs and/or comparing the performance capabilities of different exchange designs. Degradation of servic
6、e performance shall be said to exist at an exchange if the specification in Section 4 for loss-of-service does not apply, and if a proportion of the trafic offered at reference trafic loading experiences a quality of service worse than that specified under normal conditions. In some cases, quality o
7、f service under normal conditions, and quality of service under degraded conditions, cannot be distinguished. So, maximum ratios of misprocessing considered in Section 2.4, Call processing performance objectives, are applicable for processing faults, due both to sporadic errors and permanent hardwar
8、e failures. Note. This definition and structure of the quality of service is used here on a provisional basis. 2.1.1, replace with: 2.1.1 Load on Incoming Interexchange Circuits a) Reference Load A - 0.7 Erlang average occupancy on all incoming circuits with call attempts/hour/incoming circuit = 20.
9、 This figure assumes 25% ineffective call attempts. b) Reference Load B - 0.85 Erlang average occupancy on all incoming circuits with call attempts/hour/incoming circuit = 25. 2.2, replace with: 2.2 Inadequately Handled CaU Attempts 2.2.1 Definition Inadequately handled call attempts are all lost ca
10、lls occurring in the exchange under consideration (blocking as defined in CCITT Red Book 11/3), plus all call-attempts, in the handling of which, too many important delays associated with call-Set-up occur during the establishment of the connection. These are at least 3 times greater than the value
11、specified for the “0.95 probability of not exceeding” indicated in the tables under Section 2.3. Note I. For originating and transit calls, the definition of inadequately handled call attempts is assumed to apply when at least one outlet is available in the appropriate direction. Edition of Septembe
12、r 15, 1986 232b4L4 0005252 7 TICC 62-04 Page 2 2.4.2 2.4.2.1 2.4.2.2 2.4.2.3 Note 2. Provisionally, dial-tone-sending delay is not considered this context. Further study is required. Probability of inadequately handled call attempts The following table applies: 2.2.2 Type of Connection Reference Loa
13、d A Reference Load B Internal 4 x 10-2 Originating s x io- 3 x 10-2 Terminating 2 x 2 x io-* Transit I 0-3 j 2.4.2, replace with: 64 kbitls semi-permanent connections Interruption Rate A semi-permanent connection is said to be interrupted when the connection is released or when an unacceptable 64 kb
14、it/s path transmission quality across the exchange (worse than that for an alarm condition) occurs for more than Y seconds. As a long-term average, the interruption rate of a semi-permanent connection shall not be greater than X interruptions per year. Note. X and Y require further study. Release Fa
15、ilure On a provisional basis, the same specification as for switched connections (see 2.4.1.2) applies. This item requires further study. Other Failure Modes This requires further study taking into consideration: - need to recognise an interruption; - probability of an interruption; - requirements f
16、or re-establishment of interrupted connection; - any other unique requirements. 2.4.4, replace with: 2.4.4 2.4.4.1 Interruption Rate n x 64 kbitls semi-permanent connections A semi-permanent multi-slot connection is said to be interrupted, when the connection is released, or when an unacceptable tra
17、nsmission quality across the exchange (worse than that for an alarm condition) occurs for more than Y seconds on at least one 64 kbit/s path through the exchange, or when the time slot sequence integrity is no longer maintained. As a long-term average, the interruption rate of a semi-permanent conne
18、ction shall not be greater than Z interruptions per year. Note. Y and Z require further study. Y is the same value as in 2.4.2.1. To be determined. This item requires further study. 2.4.4.2 Release Failure 2.4.4.3 Other Failure Modes 9 3, insert the following: 3 EXCHANGE PERFORMANCE UNDER OVERLOAD C
19、ONDITIONS Edition of September 15, 1986 3.1 Causes of overload conditions Two groups of causes can be distinguished: a) overload by modified behaviour of subscribers (e.g. generating more call attempts because of emergency situations or altering their pattern of usage ) in comparison with that assum
20、ed for the exchange trafic engineering, causing an increase in processing load; 232b414 0005253 9 W TICS 62-04 Page 3 b) overload caused by equipment malfunction - on trunk lines, in higher (or lower) level exchanges, within the exchange itself, which causes a reduction of the available common resou
21、rces resulting in non or incorrect handling of the call attempts with the consequent phenomenon of repeated call attempts. 3.2 Definition of overload The following considerations should be taken into account when drafting that definition: One or more overload conditions of external origin should be
22、defined by CCITT. In relation to Reference loads A and B, the parameters may include: number of call attempts, call-mix, duration and repetition frequency of the overload conditions in the busy hour. The overload conditions may also be defined indirectly as the periods of time during which, with no
23、actions undertaken by the exchange, one or more of the performance objectives specified for Reference load B would not be met. Overload conditions may affect the whole exchange (general overload) or only parts of it (imbalanced/focus- sed overload). General overload may occur because the number of c
24、all attempts offered to the whole exchange reduces increasingly the processing capacity of the exchange. Imbalanced or focussed overload may occur when e.g. an abnormal number of outgoing calls are directed to a limited number of lines or trunks. In the absence of suitable control mechanisms, the on
25、set of overload results in a rapid degradation of exchange performance, which materialises as a sharp increase in process response times, leading therefore to reduced call-handling capability. Wherever, there is a mismatch between the processing capacity available and the call attempts offered to th
26、e exchange, due either to equipment malfunction or modified subscriber behaviour, it is therefore necessary to reduce the demands made on the exchange. Overload can therefore be defined as the point at which post selection delays have become so great as to cause callers to clear, anticipating a call
27、 failure. At, or just before this point, which might exceed the Reference B level of performance, a control mechanism is required. Actual behaviour and ideal behaviour of an exchange are represented by the general curve given in Figures 1 and 2 (T/CS 62-04). This curve includes two regions which are
28、 referred to as quasi linear region and non linear region. Load A point belongs to the quasi linear region. Load B point may belong to the upper part of the quasi linear region or extend marginally in the non linear region. If no actions were taken by the exchange, under overload conditions the foll
29、owing consequences might take place : - increase of the call attempts inadequately handled and of the delay probability for the call attempts - decrease of the call processing performance; - progressive decrease of the total call attempts handled by the exchange. In order to prevent and to limit the
30、se degradations, the exchange should provide the supervisions and automatic procedures (overload control) described below. The mean number of the call attempts handled in an overload condition should not be lower than that one corresponding to the system capacity in the trafic conditions assumed for
31、 the exchange engineering. For the call attempts handled in an overload condition, the exchange should maintain a performance objective as close as possible to that specified for Reference load B in Sections 2.2, 2.3 and 2.4. The extent to which the above is possible, in relation to the number of ca
32、lls offered and performance objective that should be maintained, depends on the philosophy that might be considered for handling calls. One possibility would be for the exchange to handle a minimum number of calls, e.g. corresponding to Load B, whatever the priority level of calls. Another situation
33、 could be based on a rejection of non priority calls leading ultimately to a number of handled calls which may fall below a certain level and a level of performance which may exceed that for Reference load B. This requires further study. 3.3 Behaviour of an exchange handled ; 3.4 Grade of service Ed
34、ition of September 15, 1986 m 2326414 0005254 O TJCS 62-04 Page 4 Ca 11 s / s Processed CaIls/s Processed Figure 1 (T/CS 62-04). Possible actual behaviour of an exchange with overload control. Figure 2 (T/CS 62-04). Ideal behaviour of exchange. 3.5 Overload indications Overload indications may be pr
35、ovided by: a continuous measurement of the occupancy of the resources used for call handling during short observation time (e.g. few seconds); a supervision of the filling degree of the waiting input buffers to the various call handling processes; etc. Overload indications should be given to the mai
36、ntenance staff. Edition of September 15. 1986 _ 232b414 0005255 2 TfCS 62-04 Page 5 3.6 Overload protection The control method is dependent on the particular technical arrangement of the switching system and is not subject of international Recommendations. When the exchange detects an overload condi
37、tion and before the actions specified hereafter are undertaken, the system should perform appropriate adaption of processing priority of tasks to the new situation. Different parts of the exchange may be at the same time under different operational phases; however, the actions applied to each part o
38、f the exchange should be progressively eliminated in the reverse sequence in relation to the decrease of the overload condition. For any type of switching system architecture, the exchange should avoid to route further call attempts on a particular functional block in overload conditions from other
39、functional blocks of the same exchange. In order to meet the specified performance requirements the call processing system may need to enter different operational phases as follows: Phase 1 : All seizure attempts are acknowledged and handled. The exchange can defer some or all the activities non ess
40、ential to handling the trafic offered such as administration maintenance processes (routine controls, automatic or operator handled tests on subscriber lines and circuits, etc.) and operational processes (trafic measurements and other exchange measurements, call tracings, etc.). Nevertheless the man
41、-machine communications essential for priority operational tasks should always be preserved. Phase 2: All seizure attempts are acknowledged but a limited amount handled. Priority seizures are handled. When the preliminary actions of Phase 1 are not enough to absorb the overload condition, the exchan
42、ge should adapt the trafic sources to its present call handling capacity. This is obtained by performing a discrimination of the call attempts to be handled on a priority basis. At least priority trafic should be handled. 1. On the subscriber side a specific audible indication should be sent to the
43、subscriber whenever possible, if his call attempt is not to be handled. 2. On the trunk side successive seizure of a number of idle circuits should be prevented. This can be done e.g. by blocking idle incoming circuits in the backward direction, or by sending an appropriate message to the distant ex
44、change. Phase 3: Only a limited amount of seizure attempts are acknowledged or handled. No priority treatment is given. 9 4.1, add after the last sentence: See also Recommendation T/CS 10-13 3. w 9 5, add after last sentence: Note. Co might not be a constant but depend on BHCA and traffic characteri
45、stics. REFERENCES 11 2 3 CCITT Recommendation 4.514. Performance and availability design objectives. T/CS 62-01 Recommendation. Introduction,$eld of application and basic functions for digital local and combined exchanges. T/CS 1 O- 13 Recommendation. General maintenance concept and terms. O Edition of September 15, 1986
