cancel
Showing results for 
Search instead for 
Did you mean: 

Multiproviders and Hints - RRKMULTIPROVHINT

Former Member
0 Kudos
234

Hi Dudes,

Can anybody help me with the "Multiproviders and Hints" and the table "RRKMULTIPROVHINT"?

I found a document [20 technical Tips and Tricks|http://www.comeritinc.com/UserFiles/file/tips%20tricks%20to%20speed%20%20NW%20BI%20%202009.ppt] in which the 1st optimization Tip is "Multiproviders and Hints".

I don't understand how to fill the table "RRKMULTIPROVHINT" mainly the fields POSIT & MAXOR.

Q: What are the values to fill in the table if you have a MultiCube with a logical partitionning by Time Period (2005 to 2010)?

Q: Is that optimization efficient if one has a logical partitionning by Year (for example) and one always accesses to the queries with Period and Company criteria?

Thanks in advance for your replies.

Samuel

Accepted Solutions (1)

Accepted Solutions (1)

Former Member
0 Kudos

Hello,

Does anybody have an help more effecient and concrete please?

I'd like to know how to fill (with what values) the fields POSIT and MAXOR if I add a new row into the table RRKMULTIPROVHINT.

Thank in advance for your positive feedback.

Samuel

Private_Member_101989
Contributor
0 Kudos

Samuel,

From the document:

1. Go to SE16, type in table u201CRRKMULTIPROVHINTu201D and choose New.

2. Make the entry and choose Save

But read the document.

Regards,

Jorge Diogo

Former Member
0 Kudos

Jorge Diogo

I don't ask how to create a new record in the table RRKMULTIPROVHINT but what vales must I fill in the fields POSIT & MAXOR when I create a new record.

I already knew the document that you have attached in your post and the information that I am searching are not mentioned.

Samuel

Former Member
0 Kudos

Hi Samuel,

in RRKMULTIPROVHINT you have four columns. The first column is MULTIPROV, wheras this column contains the multiprovider the hint is for (trivial) The second col is CHANM describing the partition criteria. Lets assume that you have a multiprovider "zm" with several cubes that split the data per year each and you have "zc_2009" and "zc_2010" containing data for 2009 and 2010 respectively. Within the cubes you have 0CALYEAR and 0CALMONTH as date characteristics. In RRKMULTIPROVHINT you would add two rows: one row telling that the cube might be determined (by using the hint) by evaluating the 0CALYEAR and another row to give a similar hint by specifyuing 0CALMONTH. The query logic would then try to determin if the 0CALYEAR or 0CALMONTH was restricted my a variable and prune the 2010-cube is someone selected the month 05.2009 for example.

The column POSIT in RRKMULTIPROVHINT builds an order for selecting the proper hint criteria, wheras the more precise characteristics (here 0CALMONTH) should be evaluated prior to the less ones (here 0CALYEAR).

Cheers,

Yann

Former Member
0 Kudos

Hello Yann,

Thanks for those explanation. That is exactly these kinf of answers that I wished and not some copies of SAP library...

I understood well I must add 2 records:

- ZM / 0CALENDAR / 0001 / 0

- ZM / 0CALYEAR / 0002 / 0

0001 and 0002 determining a priority, does it?

How to be sure that the queries use the Hints??? Is it possible to see it with a BW transaction or trace?

Samuel

Former Member
0 Kudos

Hi all,

yes, sorry I missed the description of the 4th column.

You may check the multiprovider explain in RSRT while testing the query. Then you can see which semantic partition was omitted.

HTH,

Yann

Answers (2)

Answers (2)

Private_Member_101989
Contributor
0 Kudos

Hello Samuel,

Check this document:

http://www.sdn.sap.com/irj/sdn/go/portal/prtroot/docs/library/uuid/751be690-0201-0010-5e80-f4f92fb4e...

For that kind of optimization, check the repartitioning (only for time partioning) functionality available in BI7.

Regards,

Jorge Diogo

former_member186445
Active Contributor
0 Kudos

actually what the doc is saying is that the standard partitioning is done by time char. you need to fill the table if you want the system to create partitions based on other char eg plant, co_area, etc...

M.