Newbie Questions, DW, Cubes, and Upgrade from 8.4 > 10.1

[cogc10]

First off I would like to apologize for these questions.  I have done alot of research prior to this and since C10 is new there is not alot of information or easy to understand information.  Below is a list of questions I have left to understand and would like to have answers, then I would have a great grasp on things.  I would extremely appreciate some answers if you are willing, Thank you so much. Here we go.

At our company, we currently have 8.4 installed. (Oracle DB) We have JD Edwards and Kronos and a couple other programs modelled with Kronos.   We don't have a Datawarehouse or cubes from my understanding or so I have been told and currently a relational Database.  We will be upgrading to 10 shortly and would like to have answers to the following to help me finish off the understand of Cognos.  We also what to have Historical Data and Predictive Analytics.

Questions:

Differences of Relational/Multi-Dimensional/Olap and Disadvantages/Advantages of each (English definitions pls)



What is a Data warehouse and do I need it? (English definitions pls)
What is a Data Cube and do I need it or both DW and Cube?(English definitions pls) Are they used together?



Data Warehouse :How do I create one? (Steps involved)
Is  a dedicated server required for a DWH?
Do I need to start from scratch from 8.4 to 10.1? 
Do I need to rewrite my Managed reports (Sophisticated reports, not the simple adhoc reports)?
Or do I need to create a Cube??  Then same questions above.
We want historical data, and want to do predictive analytics.

As per Hardware,

I'm assuming you would recommend to have a Dedicated server for Cognos,  Would a Dual Xeon with 4GB Memory be sufficient?

As well I would like to have Lotus Connections (Collaboration) and Mobile Support. I didn't see that in the C10 Demo that I installed.  Do they belong on a Separate server each?

As well for Powerplay, what is that?  I believe its for OLAP, but since it's gray to me, Do I require that, or Multi Dimensional is what I need and does it also belong on a separate server?

Again, I apologize in advance if I'm off on my assumptions above, and really do appreciate the responses.

Thank you!,
Chris

[RobsWalker68]

Hi Chris,
I can see why you have had a lot of views on this post but no answers as this is a lot of work to attempt to explain from scratch.  To try and get you some sort of answer, and shortcut the process , I've provided what I hope are useful links.

You will need to do a lot of leg work yourself and then if you have anything specific come back with questions as to be honest your questions are very generic.  Ralph Kimball has spent the last 20 years attempting to define what a Data Warehouse is and if I could explain it to you in few lines, then, well I wouldn't be writing this and  lying on a beach in Hawaii in early retirement instead.

Differences of Relational/Multi-Dimensional/Olap and Disadvantages/Advantages of each (English definitions pls)

WIKI (OLAP)
http://en.wikipedia.org/wiki/OLAP

What is a Data warehouse and do I need it? (English definitions pls)

Ralph Kimball: A Dimensional Modelling Manisfesto
http://www.ralphkimball.com/html/articles_search/articles1997/9708d15.html

Whether you need a data warehouse depends on the analysis of your user requirements and what you wish to achieve.  From the information provided I cannot tell although generally reporting requirements work well with a star schema. This could potentially be virtual as modelled in Framework Manager or physical as found in a Kimball based  DW.

Data Warehouse :How do I create one? (Steps involved)

http://www.ralphkimball.com/html/booksDWLT2.html

I'm afraid no easy shortcut on this one, either read and understand the material for your particular database vendor and implement internally with appropriately trained IS staff or employ an external Data Warehouse consultancy to do it for you.  From experience, for each subject area e.g. Sales Analysis expect to allocate 3 months per development team.  I think that gives you an idea of the potential work involved.

Is a dedicated server required for a DWH?

Yes, and normally more than one to take into account your test/development/failover environments and possibly ETL.

Do I need to start from scratch from 8.4 to 10.1? Do I need to rewrite my Managed reports (Sophisticated reports, not the simple adhoc reports)?

No, the upgrade from 8.4 to 10.1 is pretty seamless and you don't need to re-write your 8.4 content, although following best practice you should implement a test environment, migrate batches of reports and test either manually or using Lifecycle Manager.

Or do I need to create a Cube??  Then same questions above

Again depends on your particular requirements

I'm assuming you would recommend to have a Dedicated server for Cognos,  Would a Dual Xeon with 4GB Memory be sufficient?

Yes I would recommend a dedicated server, however the sizing depends on your requirements, number of users, type of reports et.c There is no hard and fast rule.  Perhaps ask your IBM Partner or IBM if you are a direct client on what thire recommendation is

As well I would like to have Lotus Connections (Collaboration) and Mobile Support. I didn't see that in the C10 Demo that I installed.  Do they belong on a Separate server each?

Lotus Connection and Cognos Go Mobile are separate downloads from the BI Server installation and Lotus Connection also requires you to implement Websphere as the application server instead of Tomcat.  You will need to consider the licensing implications as well.

I hope this has gone some way to help and point you in the right direction.

Kind Regards

Rob

[cogc10]

Thank you so much Rob,  this does point me in a direction,

Much Appreciated
Chris

[RobsWalker68]

Hi Chris,

Glad it helped.  If you have any further questions then do not hesitate to post again or if you want, PM me.

Cheers

Rob