Regression testing in Sap Business objects Data Services 4.2

dataservices job design
1

Hi Experts,

How to do Regression testing in SAP BODS 4.2? what is the Prerequisites I have to take before to do Regression testing in SAP BODS 4.2?

Thanks&Regards
Venki

ursfriend77 (BOB member since 2011-03-02)

2

That’s quite a big question. The short answer is ‘run the job before the change and record metrics, then run the job again after the change and check the metrics haven’t changed’.

There are lots of considerations to be able to give a detailed answer. Is your regression testing due to requirements changes, database changes, Data Services version changes etc.

Nemesis :australia: (BOB member since 2004-06-09)

3

Hi Nemesis,

Thanks for your swift reply

according to your question, we did some changes in source tables like column data type length increase, shuffling the Colum, and applied nullable etc… After this changes we kept ATL files in one folder, now we are going to upgrade SAP BODS version 4.0 to 4.2, After completed upgrade I have to regression testing in SAP BODS 4.2
This is my requirement.

I hope you understand what I mentioned here

please let me know if you are confuse on this requirement and let me know your valuable comments.

Thanks&regards
Venki

ursfriend77 (BOB member since 2011-03-02)

4

In that case, you’ve really got two lots of regression testing to do.

You need to make sure that the table changes haven’t caused any data issues. One way to do this is to run the job, copy the output tables to a safe location, update the tables, run the job again and compare the outputs.

Then you need to make sure that the upgrade doesn’t cause any issues. This is a bit trickier as DS version changes can have unexpected results. As an example, we upgraded to DS4.2 sp3 which has a bug that prevents pushdown in outer joins that use the ‘nvl’ function.

Nemesis :australia: (BOB member since 2004-06-09)

5

Hi Nemesis,

Thanks for your swift reply.

I need more clarification on this “we upgraded to DS4.2 sp3 which has a bug that prevents pushdown in outer joins that use the ‘nvl’ function.”

and please let me know what actions we have to take in this situation, client side upgrade completed now I have to run the jobs in test environment.

Thanks
Venki

ursfriend77 (BOB member since 2011-03-02)

6

So, we noticed that one of the jobs was taking much longer to run after the upgrade and eventually worked out that this bug was the cause. The action we took, was to find and apply a patch that fixed the issue.

Nemesis :australia: (BOB member since 2004-06-09)

7

Thank you for your swift reply and you gave wrathful information

Regards
Venki

ursfriend77 (BOB member since 2011-03-02)

8

Hi Nemesis,

I hope you are doing good.

“One way to do this is to run the job, copy the output tables to a safe location, update the tables, run the job again and compare the outputs.”

Here My doubt is how to compare the outputs? for compare i need to go database or can we compare in designer?

Thanks&Regards
Venki

ursfriend77 (BOB member since 2011-03-02)

9

I think DB Scripts is the way to go. You can use all the power and features of the specific DB to automate as much as posible the testing.

In fact, when you update versions, the testing is technical, not functional, you just need to confirm that the new output match exactly the old one. Not as easy as it is said.

For instance, use the handy HASH functions (SQL Server at least) in order to compare the outputs without any knowledge of the content of he rows.

Regards,

Andrés

aidelia_yahoo :argentina: (BOB member since 2013-07-05)

10

Hi Andres,

Thanks for your reply

if you do not mind can you say clearly how to compare the outputs? do you have any link or example please provide me it is useful to understand.

Thanks&Regards
Venki

ursfriend77 (BOB member since 2011-03-02)

11

In fact, i haven’t done systematic testing, but I did read a lot about it.

Here is a link related to what I have posted before:


Certainly you could find more best suited to your needs.

The hash technique is also useful to identify in an efficient way changes in the ETL, sadly DS doesn’t have a built-in function to make it viable with large datasets.

Regards,

Andrés

aidelia_yahoo :argentina: (BOB member since 2013-07-05)

12

Thanks Andres

Thanks
Venki

ursfriend77 (BOB member since 2011-03-02)