Relationship between BI instance and BIA (Accelerator)

[Former Member]

Hello Everyone,

There's currently one-to-one relationship between SAP Netweaver BI and a BI accelerator. Sharing multiple SAP Netweaver BI instances with a single BI accelerator is not yet supported.

I would like to find out, can we share Multiple BIA (accelerator) instances with Single SAP Netweaver BI instance?

Here is the problem I have:

I have one BI instance with about 50 cubes of data load every night. After this is done there is a BIA indexing is happening. During the indexing, the users from different countries couldn't able to run the Queries and getting timed out. But these queries are not using all the 50 cubes. These queries uses only 5 Cubes and Common Master Data.

I am trying to come up with solution how to minimize the downtime when queries are executed during the indexing.

I am thinking adding one more additional BIA and run these 5 cubes for the same BI instance. Is it possible to do that connect one BI instance to multiple BIA?

Appreciate if you can help.

Thanks,

SV

[Former Member · Accepted solution]

I don't have BIA yet, but we have been researching it a good bit. It it was my understanding that queries should fail over to the BI cube if it is not available on BIA. Isn't that happening? Or is it a case of the query performance against the BI cubes is so bad as to really not make failover an option. All teh presentations I have seen only ever show a 1 to 1 BIA to BI.

A second BIA seems like an awfully expensive solution to provide this degree of availability.

Regardless, I would pose this question to the SAP RIG Group to get a bankable answer.

[Former Member · Accepted solution]

Hi Pizzaman,

Can you please guide me how to post this question to the SAP RIG Group?

What's happening here is, after the master data is loaded in BI and then indexing will takes place in BIA. This is taking atleast three hours to complete and to do the rollup early morning between 6am to 8 am. During this time when other countries (noon time) execute the queries, its getting timed out. Because, indexing and rollup is happening at the backend. I am not sure how to solve this issue.

[Vitaliy-R · Accepted solution]

Hi,

SAP RIG is checking this forum. So, you might expect some post from them.

Regarding your scenario - something seems wrong here. Are you indexing (doing initial filling) cubes every day and this is when queries cannot be executed? Or you cannot execute queries during roll-ups? or during change run?

Regards,

-Vitaliy

[Former Member · Accepted solution]

We are indexing every because we are getting new data for both Master as well as Transactions. Master data is a full load and transactions some of them are delta loads. It takes 1.5 hours to complete the indexing in BIA. During this time, if the user access a query, it will be routed to the BI Cube and so the performance is much slower (the response time before BIA was installed) or it will be timed out. The BI Cube does not have any aggregates.

[Former Member]

Hi,

Switch of the queries using BIA for Cube by Cube ( first switch of for one cube and do index for that cube and switch on after index), same way do for all cubes.If any body runs a Report which uses the switch off Cube, its may be slow but doesn't stop the user to access Reports.

This may be a possible solution in terms of Cost ( as another BIA is more Cost)

Thanks

Sreeni