I am currently doing a project on SAP HANA using BODS as ETL tool.
I am suppossed to load large flat files - like about 50 flat files each of around 2 to 3 GB in size.
Now I am facing a starting problem of how to do Data Modeling in HANA:
1. Should I just load the flat files as I am getting - as all the calculations and Business Logics can be applied on the fly ?
2. What are some of the good standards to follow for Data Modeling in SAP HANA ?
3. What Quality Checks should I be careful when working on SAP HANA - BODS ETL ?
4. If any body can share sample Data Modeling Material it would help me to jump start ?
I have worked on BODS and Oracle Database on traditional Data warehousing projects so want to be sure that follow correct approach.
Thanks in advance for helping out.
1. Yes you can directly load the flat file in to the HANA irrespective of the size of the file. Since the HANA is an in memory computing system, it has the ability to load the numerous amount of files within the respective transaction time. You can do all the business logic within the HANA by creating the views in the content folder
2. Do the modelling as per the table structure as - segregate the master and the transaction table, since the master table is used for the Attribute view and the transaction is used for the Analytic view
3. Since you are trying to extract just the flat file, You does not need to follow any metrics. Its just the direct loading of the data from the file using the BODS. Just the Data stores are created in the ETL level to read the file and write to the HANA system
4. Do refer to the SAP HANA Developer Guide for your reference for modelling - http://help.sap.com/hana_appliance
Below are my observations.
1. It purely depends on the requirement.
For eg ,if it require transformation of trimming and substring ,it can be done on fly.But for complicated transofrmation,you can take the 3 step approach of ETL
where you first extract the data and format in a standard form in a table.and then applying the transofrmation and validation ,if any.After that you can load the data.
2. Please follow the below link for best practices in modelling of SAP HANA Data loads
3. Need to check it.
4 Pleaswe use the foloowing PDF for any reference.
Please let me know if the answer suits your query.!!
If you are only going to use flat files, you may not want to use BODS unless there is heavy transformation on the fly before writing data into HANA tables.
In general, HANA is very good at handling flat files. You have two options - 1 - to use the HANA import script and 2 - to use an XS project.
Also, let us know the full requirement so that we can guide you better.
Hope this helps.