PJI (Product Join Indicator) = CPU * 1000 divided by I/O.
UII (Unnecessary I/O Indicator) = I/O divided by (CPU * 1000).
The metrics are used to make an idea about query efficiency for consuming CPU and I/O resources.
If PJI is relatively high for a query, then the query takes many CPUs for the given number of I/Os. Maybe there is a Product Join in the Explain. Hence the name of the metric.
If UII is relatively high for a query, then it could mean that many I/O blocks are read, but relatively low number of rows actually processed. If it is a full table scan with only a few rows qualifying, then an index could reduce the I/O consumption in this case.
Both metrics are available in Viewpoint's Query Monitor portlet. And they can be calculated in DBQL by using AMPCpuTime and TotalIOCount fields.
PJI (Product Join Indicator) = CPU * 1000 divided by I/O.
UII (Unnecessary I/O Indicator) = I/O divided by (CPU * 1000).
The metrics are used to make an idea about query efficiency for consuming CPU and I/O resources.
If PJI is relatively high for a query, then the query takes many CPUs for the given number of I/Os. Maybe there is a Product Join in the Explain. Hence the name of the metric.
If UII is relatively high for a query, then it could mean that many I/O blocks are read, but relatively low number of rows actually processed. If it is a full table scan with only a few rows qualifying, then an index could reduce the I/O consumption in this case.
Both metrics are available in Viewpoint's Query Monitor portlet. And they can be calculated in DBQL by using AMPCpuTime and TotalIOCount fields.