Page 577 - Cloud computing: From paradigm to operation
P. 577

XaaS                                                     3


                                                     Appendix IV


                                            Value for response time limit
                            (This appendix does not form an integral part of this Recommendation.)


            The response time is specified in clause 7 as a general requirement for DaaS. The value of the response time-
            limit is needed in order to meet the CSC’s acceptable QoE.
            This appendix only provides general information relevant to response time. However, the response time
            described in here is not specific to cloud computing and DaaS.

            Figure IV.1 shows a response time model [b-Shneiderman].

                     User        Arrival  Server begins  Server completes  Event  User's device  User's device
                    initiates   of user's  to process  processing     result    begins       completes
                    request      request  user's request  user’s request  delivered  response  response


                        4) Transport delay  3) Server processing time  4) Transport delay  5) Render time
                                               1) Response time
                                               2) Interaction time
                                                                                            Y.3503(14)_FIV.1

                                            Figure IV.1 – Model of response time


            As shown in Figure IV.1, the response time includes transport delays, server processing time and rendering
            time on a client as defined hereafter:
            •       Response time: An interval from the moment when user initiates an activity until that user's device
                    begins to present its result on the display;
            •       Interaction time: The same as response time;
            •       Transport delay: An interval from the moment when user's request is sent until it is received by a
                    server or the time from the moment a server sends;
            •       Server processing time: Operation time used by a server to process the user's request;
            •       Rendering time: Duration for generating an image with the received event results screen on the
                    user's device.
            Values for acceptable response times may differ depending on the use case, but experimental studies show
            that a user starts to be bored or interrupts if there is no response within one second [b-Tolia]. It seems
            common that users are intolerable if the reaction time takes more than one second.
            As shown in Table IV.1 [b-Tolia], in the range from 150 ms to one second users become increasingly aware of
            the response time, and unhappy after one second.


                                   Table IV.1 – Impact of interactive response time [b-Tolia]
                             Response time                                Subjective impression

             < 150 ms                                        Crisp
             150 ms to one second                            Noticeable to annoying
             One to two seconds                              Annoying
             Two to five seconds                             Unacceptable
             Longer than five seconds                        Unusable





                                                                                                         569
   572   573   574   575   576   577   578   579   580   581   582