It seems that for computing the storage space of a SE, VAPOR taks into account the attributes GLUE2StorageShareCapacity(TotalSize, FreeSize, UsedSize). The storage areas can overlap each other, and in many cases this brings to an overestimation of the storage capacity.
So the most proper attributes to consider would the GLUE2StorageServiceCapacity(TotalSize, FreeSize, UsedSize) ones.
The benchmark value (either hep-spec or SI2K) refers to a CE, so in case of a site with more than one CE, it is not correct summing the CE hep-spec (or SI2K) values for getting a global site hep-spec value (and so it is not correct a NGI global hep-spec value).
So in the summary tables (https://operations-portal.egi.eu/vapor/resources/GL2ResSummary ), for each site, instead of displaying the global benchmark values, it could be useful having the "computation power":
for each CE, compute the value: hep-spec * LogicalCPUs
then sum the several CE computation power quantities for getting the total site computation power.
hence for having the NGI computation power you can simply sum over the single site values.
When a site provide 2 or more CEs with the same amount of Physical and Logical CPUs, VAPOR should take into account only one of them for computing the total CPUs values of the site.
