Why ABAP code doesnt work in SAP HANA in memory?

[Former Member]

Hi All,

SAP HANA is not my area and I just started learning basics for SAP HANA.

Can anybody tell me why ABAP code cant run or support in SAP HANA in memory?

I was working with SAP BW IP integrated planning functionality in which SAP BW is build on top of the HANA and in this IP while creating any planning functions I need to write FOX code and cant use ABAP coding. One of the reason of using FOX code is for performance this I came to konw. But I am still not clear why FOX will work more faster in SAP HANA database than ABAP.

Thanks in Advance.

[brianbernard]

Hi all,

I just want to point out that there is a new space dedicated to ABAP for HANA:

http://scn.sap.com/community/abap-for-hana

Cheers,

Brian

[Former Member]

Hi Brian,

Thanks for sharing the new space. I have always thought about this and I already am having tonnes of question on the execution part.

I'll attend the TechEd sessions and see where it is going. Way to go!

/bikas    

[former_member184768]

Hi Mavoo,

I am sure the experts have provided you all the required and valuable information.

But just want to mention that you brought up a very important point. This is something, I am sure, most of us are waiting for. In a different test case compared to IP, like data load in BW on HANA, we experience that the extraction, writing of data to DSO and activation is very fast as compared the traditional BW on Non-HANA database, but the ABAP routine execution time does not show that high performance improvement. This is due the very fact explained above and what you mentioned that ABAP execution doesn't happen in HANA.

We also looked into the alternatives of executing HANA stored procedures from ABAP code, but it is a work around. Although there is enough documentation available on how to achieve it, it is still a work around.

I am very much looking forward to attending the session in TechEd to see how ABAP and HANA are getting integrated.

Thanks for bringing up this discussion. It is very helpful.

Regards,

Ravi

[schneidertho]

Hi Mavoo,

I like to add a few more thoughts (even though I am not a BW specialist).

1. we are currenty looking into possibilities to better integrate ABAP + SAP HANA and to optimize our ABAP application server for running on top of SAP HANA. In case you are at TechEd, don't miss TEC106 and/or CD202
2. when you look at ABAP coding, this is executed on the application server (except the SQL statements which are send to the underlying database, be it SAP HANA or a traditional RDBMS)
3. it is not always the best to do complex calculations on the application server instead of doing them on the database server (especially if the database = SAP HANA)
4. when you make use of the planning capabilities BW offers you (instead of writing ABAP code), you can push down planning functions to the SAP HANA database by means of the planning applications kit, see help.sap.com and SAP note 1637199

Best regards

Thorsten

[former_member191765]

Hello Thorsten & Ravi,

Can you please explain how COPA reporting working(After COPA accelerator implementation) fast though it uses ABAP for execution.?

Thank you,

Anjan...

[schneidertho]

Hi Anjan,

to be honest I don't know the specifics of the CO-PA accelerator, but let me try to answer your question in more general terms (because basically all accelerators rely on this pattern).

The general idea behind an accelerator is that you have an ABAP program which is slow.

The slowness of this program should be mainly caused by database access. Maybe the database access causes lots of I/O load (because the traditional database might not have all the data in the buffer). The program might look like this (I know... life is not always that simple):

DO SOMETHING 1.

SELECT ... FROM... INTO...

DO SOMETHING 2.

DO SOMETHING 3.

DO SOMETHING 4.

Now imagine that you send the SELECT statement not against your traditional database, but against SAP HANA: the data is in main memory (using column store + compression). That might speed up your program significantly.

So you could say the general idea behind the accelerator pattern is: keep the as much of the ABAP program as possible and make some dedicated changes to 'push down' application logic to the database layer where it can be performed faster. That might also include using the standard Open SQL aggregation functions instead of aggregating data in ABAP.

What happens to ABAP programs which are slow, but the slowness is caused by logic on the application server? Can such ABAP programs also benefit from SAP HANA? Basically yes. But the acceleration might not be as straight forward as above, because you will have to think more about how you can achieve a 'push down' of application logic and this might require some refactoring of your program.

Hope that helps a bit.

Best regards

Thorsten

[former_member93238]

Hi Mavoo,

First, SAP HANA is mainly a database.

second,when talking about ABAP and HANA we need to think more about integration and leveraging.

below is a very good article and a previous discussion  .It will give you better view and outlook.

http://scn.sap.com/docs/DOC-29357

http://scn.sap.com/thread/3202965

Hope this helps.

Moussa