Page 22 - Methodology for inter-operator and cross-border P2P money transfers
P. 22

11.3.1   Team Naming                               11�4 Naming and formatting conventions for data
            For data processing, consistent naming over all    objects
            data sources of primary importance as it is used to   In order to facilitate efficient and error-free trans-
            combine data, and errors will lead to incorrect KPI   fer of data between file media and databases, the
            evaluation.                                        following conventions must be met:
            The following assumptions/prescriptions are made:
                                                               •  Column names in data files (e.g., Excel®) shall only
            •  A DFS operator is identified by, in this order, the   contain alphanumeric characters; shall not begin
               network name and the country name. Examples:      with a numeric character, and shall not contain
               MTN Rwanda; Airtel Uganda; Vodafone Ghana.        whitespace characters.
            •  Team names shall be chosen accordingly, i.e., the   •  Date and time shall be given in one of the stan-
               team name should be the same as the operator      dard formats. Formats shall not be mixed within
               name.                                             the same column.
            •  A team operates a “logical group” of one DFS    •  In particular for time information, data integrity
               device and one Observer device (see Test scenar-  has to be maintained over the whole processing
               io under consideration).                          chain. In particular, this applies to decimal point/
            •  There are only two permitted types of device/     comma and delimiters in general. Also, it is known
               function allocation:                              that millisecond formats can be troublesome. It
                                                                 is highly recommended to run tests before going
                a)  One device is fixedly assigned to the DFS    into full-size data processing.
                   role (i.e., is set up for a particular DFS opera-  •  A proven method to enhance conversion robust-
                   tor), and the second device is fixedly set up   ness is using “decorated” elements. For instance,
                   for the Observer role.                        if a timestamp shall be given with milliseconds, in
                b)  Both devices are set up for both roles, but the   can be written with a preceding fixed-size string.
                   DFS operator on both devices is the same.     In that case, it will be imported as a string (e.g.,
                                                                 nvarchar) to the database. Once in the database,
               If configuration b) is used, data taken by the MSW   substring/cast/parsing operations can be used to
               and network performance background testing are    convert this element to a datetime element again
               treated as equivalent with respect to the generat-  without the risk of loss of information.
               ing devices.

            The word “team” has historic origins. In consequence,   11�5  Local log sheets
            the team name represents, and stands for, a partic-  There are two variants of log sheets. The FTL assess-
            ular DFS operator. In this sense “team” is a logical   es the given situation and decides which variant is
            entity rather than meaning a group of people.      to be used, and which information is pre-printed and
            Therefore,  a  team  can  be  represented  by  a  single   which has to be entered by the teams. This decision
            person. It is possible that a single person or a group   is based on the level of skill and experience of teams,
            of persons operates multiple devices, or switches   and expected frame conditions in a given location.
            between different set-ups. In that case, the docu-
            mentation described in subsequent sections (TAL)   •  A rather explicit Location Log Sheet which has
            needs to list each configuration where the DFS oper-  fields for basic scenario and set-up related infor-
            ator changes as a separate team with the respective   mation, and provides a detailed check list for
            time window information.                             elements to be checked periodically (on a prox.
            This is because there must be absolute clarity (i.e.,   2-hourly schedule).
            at all times) in the relation between the Observer   •  A short-form Session/Location Log Sheet which
            device/data ID’s (multistopwatch and background      also collects information essential for measure-
            testing) and the DFS operator under test.            ment data allocation, but has a simplified status
                                                                 check section which just asks for a general con-
                                                                 firmation that operating conditions are still within
                                                                 valid parameters.

                                                               It  is  good  practice  to  use actual file names which
                                                               contain a reference to the project or campaign they



           20    Methodology for inter-operator and cross-border P2P money transfers
   17   18   19   20   21   22   23   24   25   26   27