HANA EIM Cleansing Data Footprint

[shyam_uthaman]

Hi All,

I am part of a project which is considering using EIM for doing the transformation logic instead of the traditional calculation views.

My concerns are:

1. Aren't we doubling the data footprint by again storing the result in a persistent table?

2. For reporting purposes, will we still create a view on the resultant table?

Please let me know your views on this.

Thanks,

Shyam

[lbreddemann · Accepted solution]

Hi Shyam,

not too sure what you're asking...

EIM provides ETL functionality integrated into SAP HANA.

So for usage scenarios where you want to get data from external sources into your SAP HANA instance, including the required cleansing & transformation, this might be more fitting than installing a separate ETL system like SAP DataServices.

WIth EIM you've everything within your SAP HANA and don't require additional servers etc.

Now, does that change how much data you store?

Not necessarily. You still get the same data into SAP HANA than with any other solution.

The idea is not to replace your data transformation you do on top of the data that is already in your instance. So, no, I don't see that this replaces your transformation views.

And the same angle applies to the second question.

If you cleverly build your reporting views as a separate layer (so that these are not affected every time you change the big amount of underlying view logic) then you should not even have the need to touch those.

Reporting views are really the report-facing data structure of your solution - tables are not.

- Lars

[shyam_uthaman · Accepted solution]

Hi Lars,

Thank you for the reply.

Lets say for a scenario where I have to join 2 tables A and B, I assumed that I can now use a Flowgraph to do it provided the data has already been provisioned into HANA by SLT or otherwise.

Or is it that like dataservices,I can connect to the source system, say ECC and create a flowgraph, without pre-replicating data/metadata for tables A and B and when I execute that flowgraph, the data passes through that logic and the final result only comes and sits in HANA, not the source tables?

That is what I meant with the original question. If I have source tables A and B plus now a join result table C instead of an analytical/calc view for C, my net memory taken is A+B+C

But if EIM doesnt require source tables to be in HANA, the net memory required will be only the result table C.

Please let me know.

Thanks,

Shyam

[Former Member · Accepted solution]

Hi Shyam,

You can have two scenarios with EIM in HANA.

1) Moving data with in HANA (one schema to other schema or within same schema).

Lets say you are already replicating data from source to HANA using SLT/DS. You have data physically stored in HANA. If you want to join join two tables using EIM then yes you are going store the once again in the system. Now your system memory is as you said A+B+C.

2) Loading data from source to HANA.

In this scenario, you have remote source configured either using SDA or Generic Adapters. now you can access the source tables virtually if you want to load data to HANA then you can do using EIM. now if you want to join two virtual tables and load the result into HANA, then you are storing only result  in HANA.

please check the below image.

Regards,

Venkat N.

[shyam_uthaman · Accepted solution]

Hi

That clears up a lot.

Thanks for that. Also, I would like to understand, that if i have scenario 1:

1. Can I replace view based modeling strategy with EIM flowgraphs instead? What would be a best practice as per SP9? i personally do not like the data duplication that goes on. But there has been a strong push for this strategy and I would like to understand the Pros of using this and if this can to an extent replace the Calc/Analytical View based modeling.

2. Can Analytical/Calc views be used as an input to EIM flowgraphs?

Regards,

Shyam

[Former Member · Accepted solution]

Hi Shyam,

1) My personal opinion about using EIM for models in HANA would be NO. I would rather prefer to have EIM to bring data from source to HANA.

2) I think you can use models also as source in EIM (need to check in the system and confirm you).

Regards,

Venkat N

[Former Member · Accepted solution]

Hi Shyam,

You can use modeling objects as source in EIM.

Regards,

Venkat N

[shyam_uthaman · Accepted solution]

Hi Venkat,

Thanks a lot for the responses.

I have much more clarity on this now.

I am marking this thread as answered but if anyone has anything that contradicts the above statements, please feel free to add.

Thanks also for your response.

Regards,

Shyam

[shyam_uthaman]

.. any thoughts?