New to OBIEE - High level steps

Question

Hi All,

I am not sure if it's the correct subspace to post my question.

If it's not, just direct me to the correct one and i'll move there

I am new to OBIEE and trying to understand the high level installation steps.

We use an old OBIEE - OBIA version (11.1.1.6) with Informatica / DAC for our ETL / Scheduling.

To better understand OBIEE, I am doing the installation on my lab but I am unsure about the high level steps.

I am following this Oracle Doc (Link) but it is strickly for the OBIEE part. No references to OBIA / Informatica / DAC.

Is there any order in which Informatica / DAC / OBIEE / OBIA should be install?

For example, for now, I have a data warehouse DB install with the MDS / BIPLATFORM schema created (with RCU) and OBIEE installed (choose Enterprise Install).

Can I simply proceed with Informatica / DAC then OBIA?

Is there any Oracle DOC that describe these high level steps?

Thank you,

Félix

[Deleted User] · Answer

Nayas wrote:
- OBIEE should be "light", as it "only" read data that was populated by INFA in the DWH database to populate/generate the dashboards / reports. (I assume the bigger the table, the more heavy it will be)

Not really. If there are many factors:

- how many users (concurrent)

- how many queries (concurrent9
- how good or bad is the physical data model and its optomization

- how good or bad is the logical model and its optimization

- how good or bad are the analyses and dashboards

Nayas wrote:- Informatica will use heavy on I/O as it generate files in it's cache, from the source dataset and iterate through every transformation, until it pushes the last dataset to DWH database.- DWH will be heavy on IO for obvious reasonsIn my scenario, OLTP / DWH databases sitting on srv1 are not the sources databases.OLTP holds the informatica / dac repository metadata and dwh is the target data warehouse database.Sources are sitting on other servers.

Ok so this sounds to me that it's a demo / poc / lab environment? In that case...yeah it maybe would make sense to have this kind of split.

Nayas wrote:Also, do I understand correctly that I need to install OBIEE on my personnal laptop (running windows) and install OBIA on my laptop ?If so, can I simply install obiee 11.1.1.7 on my laptop, install obia 7.9.6.4 instead of installing obiee 11.1.1.6 as I run on my Unix installation?

OBIEE can run on Linux as well - it actually runs a lot better then on Windows ;-) Since "OBIA" is just content you can even put OBIEE 12c on top of it as OBIEE doesn't care what its sources are.

Nayas · Answer

Hi Christian,

First of all, thank to for the time to answer my questions !

Would you mind to explain a bit more, what are the heavy component?

From my understanding :
- OBIEE should be "light", as it "only" read data that was populated by INFA in the DWH database to populate/generate the dashboards / reports. (I assume the bigger the table, the more heavy it will be)

- Informatica will use heavy on I/O as it generate files in it's cache, from the source dataset and iterate through every transformation, until it pushes the last dataset to DWH database.

- DWH will be heavy on IO for obvious reasons

In my scenario, OLTP / DWH databases sitting on srv1 are not the sources databases.

OLTP holds the informatica / dac repository metadata and dwh is the target data warehouse database.

Sources are sitting on other servers.

Also, do I understand correctly that I need to install OBIEE on my personnal laptop (running windows) and install OBIA on my laptop ?

If so, can I simply install obiee 11.1.1.7 on my laptop, install obia 7.9.6.4 instead of installing obiee 11.1.1.6 as I run on my Unix installation?

Have a great day,

Félix

[Deleted User] · Answer

Hi felix,

Nayas wrote:I am not sure to understand why is OBIA.Are they pre-built Oracle ETL for Informatica for some sources?

Yes, OBIA are pre-built ETL routines and data models for products like Oracle eBusiness, Siebel CRM, Peoplesoft etc.

Nayas wrote:
I have 2 unix server :unxsrv1 : 2 rdbms databases (oltp / dwh) and obieeunxsrv2 : informatica / dac

That's pretty much the least performant option :-)

Informatica is used in an "ETL" - extract, transform, load - manner and data has to physically flow over the Informatica server, so if you do it like this you will move the data from server 1 to server 2 and then back to server 1 for...not much.

If you put Informatica alone on server 2, then you put the least used components on server 2 whereas all the heavy things are in server 1.