Page 34 - Methodology for measurement of Quality of Service (QoS) Key Performance Performance Indicators (KPIs) for Digital Financial Services
P. 34

ANNEX D
           KPI basics








           D.1 OVERVIEW                                         Each phase is marked by exactly two events. The
                                                              starting event (E1) stands for an attempt to reach the
           The KPI defined in this clause have been introduced   functional goal associated with this phase.
           in  the  Focus  Group  Technical  Report  QoS  and  QoE   A successful completion of a phase is indicated by an
           aspects of Digital Financial Services (05/2016) by the   event E2 (positive success). If the phase has not been
           ITU-T Focus Group Digital Financial Services.
                                                              completed successfully, this is either indicated by an ex-
                                                              plicit event, or by reaching a given time-out condition.
           D.2 TERMINOLOGY                                    Every phase must have such a time-out condition as pa-
                                                              rameter. This ensures that no undefined state occurs,
           The definitions for computation of KPI are based on the
           formal event codes defined in clause 5.            ■  NOTE 1: Without such a time-out condition, an explic-
             The following is, for the purpose of easy reading, a   it list of events indicating failure would be required. In
           condensed description of the model developed and ex-  case the system under test yields a response previ-
           plained in detail in ETSI TS 102 250-2  (or Rec. ITU-T   ously unknown or not caught by a definition, the test
                                             3
           E.804 ).                                             case state would be undefined. With the requirement
                4
             A transaction is defined to be a single instance of   of a positive success indicator and a catch-all time-
           a particular use case. Each transaction produces a se-  out condition, this case is prevented.
           quence of events. From these events, the properties for   ■  NOTE  2:  Instead  of  an  explicit  phase-wise  time-out
           the respective transaction are derived. The minimum set   condition, there may be a ‘global’ time-out, with the
           of information includes the result of the transaction (in   clock started at the very beginning of the transaction.
           the simplest case, success or failure; other result types
           can also occur).                                   Each event is associated with a time stamp. Therefore,
             Typically in case of success, the elapsed time (dura-  an event has two basic properties: Its presence (i.e. if a
           tion of the transaction) can also be computed; Depend-  particular event is present within a given transaction),
           ing on the nature and structure of a transaction, such   and, if it is present, its time stamp.
           computations may be complex and involve a multitude   ■  NOTE: Events are understood to be logical events,
           of events.                                           which are assumed to be unique by definition. These
             A transaction can consist of several phases. Typical-
           ly, a transaction has a layered or hierarchical structure,   logical events are set by actual technical input, such
                                                                as  occurrence of particular protocol  messages  or
           with a single-phase description at the top representing   data items at some point of observation. Such tech-
           the ‘end to end’ view.                               nical events are not necessarily unique. In such cases,
             Each such layer of description is called a view on the
           transaction, with the degree of detail, i.e. the number of   a state logic is assumed to exist which produces re-
                                                                spective logical events from technical input.
           phases, usually increasing with lower-level views. In a lower-
           level view, phases may be partitions of respective higher-   Example: Assume two events E1 and E2, where E1 stands
           level phases, or phase boundaries may have a different   for an attempt to achieve a given functional goal (‘try’),
           phase composition.                                 and E2 indicates that this goal has been reached (‘suc-
             In the QoS context, a phase is usually related to a   cess’):
           part of an overall transaction which has meaning in a
           user-perception way and is connected to events which
           are observable by a typical user. In a more diagnos-  D.3 EXPRESSIONS
           tics-related context, however, phases can also be relat-  For KPI computation, the  following expressions are
           ed to protocol messages or other events which are not   defined. For actual KPI definition, respective Ex are re-
           typically user-visible.                            placed by their trigger point ID’s (see also Event Codes).






           32 • Methodology for measurement of Quality of Service (QoS) Key Performance Indicators (KPIs) for Digital Financial Services
   29   30   31   32   33   34   35   36   37   38   39