Performance Question - Precalculation vs expression & Server Sizing

Thu, 22 Sep 2011 18:38:39 GMT

Hello,

We are in the early stage of our QV deployment .

I have created an app where the underlying table has approximately 25Million records.

At this time we are on a very basic server 2 Socket single core CPU with 16GB ram.

(We are in the processes of obtaining a new server )

The current application has 18 tabs and planning to add two more. There are several calculated dimensions and expressions.

Switching between the most complex tabs take approximately 30 seconds to render.

We have few of options in front of us. I would like get your expert opinion on those.

1.Pre-calculate (in QV load Script ) some of the expression/Calc-Dim

· Expected result - Faster dashboard rendering

· Drawback - Losses the flexibility of on-the-fly updating and any changes need reloading of the script

2. Pre-Aggregate some of the facts (I am NOT for this)

· Expected Result - faster app

· Drawback - Duplicate facts may present and the report could be inaccurate

3. Get a better server

· Expected Result - Super fast application

· Drawback - Budget

I have a very little understanding on QV server sizing - need your help here

We are expecting another 5 more similarly configured application on the same server

4. Break the app in to many

· Expected result : Smaller app may run faster

· Drawback : more app means more licensing cost and user has to navigate thru multiple apps

My Questions are

· What is the recommended/common sense solution to make an app run faster?

· What would be a good server for our needs?

· Does multi threaded cpu helps at all(Does QV support multi threading)? or more core make better sense? (we can get a multi core AMD for cheaper)

Thanks in advance,

Aji Paul.

topic Performance Question - Precalculation vs expression & Server Sizing in QlikView