Skip to Content
2

HANA DI (HDI) and its Impact

Oct 11, 2016 at 03:46 PM

326

avatar image

Hello All,

With HANA SPS 11 and 12 SAP Introduced XSA and with it the concept of container(HDI). I am trying wrap my head around this concept and what it means HANA Modeling in future.
I believe there are customers who are using HANA primarily as Data warehouse solution where in we define schemas, design tables using HDBDD, then use different data provisioning tools to populate these tables, build views,secure views using analytical privileges (attribute,analytical and calculation views) and finally consume these views in BI Reporting tools like BOBJ,Qlikview,Tableau.

I would like to understand will SAP continue to support this current(above) modeling approach or we need to follow the new modeling approach of using HDI containers even though we don't intend to us any of XS/XSA features.

I would really appreciate some clarity regarding this.

Thanks,

Shivkumar

xsa | hdi
10 |10000 characters needed characters left characters exceeded
* Please Login or Register to Answer, Follow or Comment.

2 Answers

Thomas Jung
Oct 11, 2016 at 06:55 PM
1

HDI is not restricted to XS/XSA. Its many benefits are also extended to database level development as well.

As far as if you can continue to develop the old way - the answer is not for too much longer. The HANA repository will go away and artifacts like HDBDD must be switched over to HDI. So essentially the design time abstraction artifacts are tied to HDI.

If you wish to continue to use schemas directly, you can still use SQL and DDL directly to define those objects. But you give up many of the lifecycle management advantages that HDI (like the repository before it) could provide. And of course if you plan to use Calculation Views, they are always design time based and must be moved to HDI at some point.

Show 2 Share
10 |10000 characters needed characters left characters exceeded

Hello Thomas,

Thanks for taking time and responding to my question. I have few follow up questions (I am asking purely from HANA as data warehousing platform perspective)

1) How long will be the HANA Repository supported ?

2) Today we use analytical privileges/roles to secure data and HANA views. Will this change in HDI world and how ?

3) Today user authenticate against HANA via their HANA database users (we use SAML and kerberos SSO where in the AD accounts are tied to their respective database users ). Will this change in HDI world and how ?

4) What does this mean to reporting tools like (BOBJ , Qlikview , Tableau)? Today most of the tools have native connectivity to HANA and look for HANA Views under _SYS_BIC schema. Will this change in HDI world and how ?

I am really curious to understand how this architecture change (HDI) will impact HANA Modeling in future. Especially for customers who use HANA as data warehousing solution ?

0

Hello Thomas,

Thank you for taking time to clarify things.
I have few follow up questions (that I am asking purely from HANA as data warehousing solution perspective)

How soon will HANA Repository go away?

Security :
Today we use analytical privileges/roles to secure the data and views.How will this change in HDI world?

Authentication :
Today we have AD users tied to their individual HANA database users via SAML SSO. Will this change in HDI world ?

What does this mean to reporting tools (BOBJ , Tableau , Qlikview) that connect to HANA via _SYS_BIC ?


0
Shiv uppin Oct 13, 2016 at 01:48 PM
1

Hello Thomas,

Thanks for taking time and responding to my question. I have few follow up questions (I am asking purely from HANA as data warehousing platform perspective)

1) How long will be the HANA Repository supported ?

2) Today we use analytical privileges/roles to secure data and HANA views. Will this change in HDI world and how ?

3) Today user authenticate against HANA via their HANA database users (we use SAML and kerberos SSO where in the AD accounts are tied to their respective database users ). Will this change in HDI world and how ?

4) What does this mean to reporting tools like (BOBJ , Qlikview , Tableau)? Today most of the tools have native connectivity to HANA and look for HANA Views under _SYS_BIC schema. Will this change in HDI world and how ?

I am really curious to understand how this architecture change (HDI) will impact HANA Modeling in future. Especially for customers who use HANA as data warehousing solution ?

Show 3 Share
10 |10000 characters needed characters left characters exceeded

>1) How long will be the HANA Repository supported ?

It hasn't been decided yet when support will be removed. The next release of HANA scheduled for November still has it there.

>2) Today we use analytical privileges/roles to secure data and HANA views. Will this change in HDI world and how ?

Analytical privileges will need to be converted from the XML to SQL format. But that isn't that big a deal. The bigger impact is the change to the overall security model concepts of containerization in HDI.

>3) Today user authenticate against HANA via their HANA database users (we use SAML and kerberos SSO where in the AD accounts are tied to their respective database users ). Will this change in HDI world and how ?

Most likely yes. HDI and XSA are designed so that external authentication is now the norm. Database users aren't application users by default and we use technical users to connect to the database from the application layer. This doesn't mean that you can't setup application users as database users. You can use the HANA DB user store as your SAML "external" authentication. But many customers complained about having to do exactly that - setup all their end users in the DB. The new concepts we introduce make it so that is no longer a requirement or even the default approach.

>4) What does this mean to reporting tools like (BOBJ , Qlikview , Tableau)?

Quite a bit will change. There is no _SYS_REPO. Calc Views are no longer stored in _SYS_BIC. So many assumptions these tools made about where to find Calc Views and how to execute them are no longer true. Until those tools are adjusted to understand HDI containers, you might find that you have to use direct JDBC or ODBC connectivity options instead of HANA specific ones.

>Today most of the tools have native connectivity to HANA and look for HANA Views under _SYS_BIC schema. Will this change in HDI world and how ?

Yes this does change. Calculation Views now live in their application specific schemas. No more single central schema for them. This is good in the long run for security, control, etc; but certainly will have some growing pains associated with such a considerable change.

2

Thank you Thomas.

> you might find that you have to use direct JDBC or ODBC connectivity options instead of HANA specific ones.

This will involve using HANA database user for connectivity or technical user ?

In future all the Front end BI applications (BOBJ, tableau , Qlikview) will they connect to HANA (views in HDI container) using database user or technical users ?

0

When you connect via JDBC/ODBC in 3rd party reporting tools you will most likely be using database user. These exporting reporting tools don't yet understand HDI containers.

0