Page Comparison

The number of queries (reporting usage) for whole InfoProviders are InfoProvider is calculated as summation sum of query executions that accessed this an InfoProvider. 

The main source of this information can be found is stored in the /DVD/QS tables where , which contain the collected extended query statistics (extended) collected from a system resides. This information is gathered by Datavard Insights query statistics enhancement which is part of Datavard Insights product transport. Number of queries (reporting usage) . The number of queries that were run on an InfoProvider data part is calculated as follows:

• For each query that was run on InfoProvider directly:
  • Characteristic selections are checked and for all data parts of InfoProvider only those that belong in the selection The system checks a selection of characteristics and only those InfoProvider data parts that belong to this selection, are marked as read by the query.
  • If no relevant characteristic selection was usedthe time characteristic wasn't found, all data parts are marked as read by the a query.
  • For each query that was run on MultiproviderMultiProvider, SPO, Compounded compound InfoProvider:
    • If standard query statistics information (OLAP) is available for a query and a query didn’t use cache, only InfoProviders that are specified as partial InfoProvider are child InfoProviders of the above mentioned InfoProviders, are processed by next steps and can be found in the results.
    • Further checks are performed if If the 0INFOPROV selection was used in the query id. If yes , then only child InfoProviders that belong to this selection are processed. Further processing is the same as for direct inforpovider query read.

...

• • • by a query that was run on InfoProvider directly.

You can customize, whether you want to calculate the weighted average for of data groups that include more values (e.g. data group <=2013 includes (2013, 2012,..)) or do simple summation for perform a simple sum of them, that which is used by default.

In the following example are listed reasons, why you can find reason why you could use weight weighted average for InfoProvider data groups that includes more values. 

Example:

We have The InfoProvider: CO_AB that was analyzed by using the 0CALYEAR characteristic. As a result InfoProvider data were distributed was divided into the following values:

1. Data with 0CALYEAR = 2014, size = 5 GB, reporting usage = 150 
2. Data with 0CALYEAR = 2013, size = 15 GB, reporting usage = 85
3. Data with 0CALYEAR = 2012, size = 18 GB, reporting usage = 60 
4. Data with 0CALYEAR = 2011, size = 14 GB, reporting usage = 37 
5. Data with 0CALYEAR = 2010, size = 563 GB, reporting usage = 2 

We are using STANDARD data In this example is displayed a Data distribution analysis that looks followingas follows:

>=2018, 2017, 2016, 2015, 2014, <=2013

In our example the total number of queries executed on top of CO_AB InfoProvider was is 162. If When we use calculate the sum of reporting usage queries of data parts this would give us for for the data group <=2013, we would get the value 184. As each query can access 0..N data parts of InfoProvider. If we use the weighted average, we will get obtain the following number.

w.a. (( 15 * 85 ) + ( 18 * 60 ) + ( 14 * 37 ) + ( 563 * 2 )) / ( 15 + 18 + 14 + 563 ) = 6.556

Value ~6 means that data group <=2013 of the InfoProvider data have has a really low reporting usage, so the potential of this InfoProvider for archiving to be archived is high. 

...

In new Data Distribution analyses weight average is used for calculation of data groups reporting usage (as they can cumulate multiple data parts) while InfoProvider total usage is analyzed separately.

How to switch on weighted average

You have can choose between the option to switch between calculation of weighted average and summation for perform the calculation based on weighted average or based on the sum of InfoProvider data parts. By default summation is used the system uses the sum, but you can switch it change this in expert settings.

Go to Enter the transaction /DVD/RL_SETT_EXPERT and fill in the the tool name HM_BW and execute.

Image Modified

Change If you want to use the weighted average, change the parameter HM_BW_DATA_GROUP_AGG_WEIGH_AVG to X if you want to use weighted average.

Image Modified

Results will are going to be recalculated after next data distribution execution of the collector execution.. (The Monitoring profile HMBW_1DAY)

Why the number of InfoProvider usage can be higher than

...

sum of it's data

...

groups

Some times Sometimes it can happen situation where that we have obtain the following result recordrecord as a result of the Data distribution analysis:

We can see that the size was successfully distributes distributed into data parts groups (Split time characteristic was identified) but . However, the usage for the whole InfoProvider is 102, question is why but these 102 usages wasnweren't distributed to InfoProvider's data partsgroups. This can have multiple several reasons:

Reason 1: Empty data parts groups cannot have any usage

What we can say about this InfoProvider is that it have data only from data part This InfoProvider has data only from the data groups 2014 and <=2013. Therefore possible usage of data could be found added only in those to these two data parts no othersgroups.

Note that we We are analyzing only online data size, therefore if the particular data part was group has been already archived, the size for part this group will be 0 in result.

Reason 2: Selections / filters that was were used during the execution of a query execution contain a filter for a time characteristic but it point which points to an empty data partgroup.

Example:

Lets have query execution with following filters for particular An execution of a query with the following selections for the above InfoProvider:

What we can read see from the above filters, is that user try tried to read data from 2016, but the InfoProvider ZCUBE_XY doesn't have any data in 2016 data partgroup, therefore we cannot add usage therethe usage has the value 0 in the record. If you are wondering why want to know, why the object ZMONTH wasn't use read following documentationused, you can check the section Additional Time Characteristics.