aggregates are suset or smaller cubes, it also contains dimensions and fact tables.we will aggragate chars which are in drill down of perticular query.Ex: if cube is having 10 records- if i build aggregate with certain chars, records sholud be aggregated to 1 record.factcount will be coming automatically when you crate aggregates.it will tell u about how many records are aggregated to form one record.LISTCUBE is the t.code to check aggregates.when ever will run the query first it will read the aggregates, so for performance issues we will create aggregates on perticular chars.
when u load data to cube , data comes from F.Table , after compressing it moves to E.Table.Compression is nothing but deleting the request id.compression is used to aggregate the same char combination records which are stored in different request.aggregate same char values stored in different requests move the data F.Table to E table.deletes the f.table content.after compressing, we cant delete data based on request id.if we get any wrong records in request and request is compressed then we cant delete the request, we will reverse posting fron PSA and compress.