Project

General

Profile

Wiki » Description of the global mechanism »

Aggregation of data and computation with the different methods

Aggregation of Data per hour

For a given timestamp (one hour) all services of a VO / NGI are aggregated depending from their service type .

<timestamp service_type='CE'  key="2012-03-14T08:00:00Z">
    <service   name =”service1” availability="1.0" reliability="1.0" maintenance="0.0"/>
    <service name =”service2” availability="0.8" reliability="0.8" maintenance="0.0"/>
    […]
</timestamp>

<timestamp service_type='SE'  key="2012-03-14T08:00:00Z">
    <service   name =”service3” availability="1.0" reliability="1.0" maintenance="0.0"/>
    <service name =”service4” availability="0.7" reliability="0.8" maintenance="0.1"/>
    […]
</timestamp>

Computation

So when the data are aggregated per hour and per service type for the whole list of service (All CE for a given VO , all SE ...) .
We apply different algorithms on this hour on the group of services :

  • the "Oring" method : if a service in the list of service is available , the whole group of service is available
  • the average method : the availability is given as average value of all availabilities
  • the threshold method (70%,80% and 90%) : if the percentage of available services (av=1 or re=1) in the list is greater than the specified threshold then the availability of the group is equal to one otherwise equal to 0 .