synthetic key

[deepakqlikview_123]

Is synthetic keys gives faulty result.

Thanks

[anbu1984]

No

[tresesco]

Read this, one of the best threads in this community: Should We Stop Worrying and Love the Synthetic Key?

[hic · Accepted solution]

No, synthetic keys are not bad - not in themselves. And they don't give faulty results.

But if you have an incorrect data model, you will end up with many synthetic keys. And faulty results. But the faulty results are a result of the incorrect data model, not of the synthetic keys.

Read this

http://community.qlik.com/blogs/qlikviewdesignblog/2013/04/16/synthetic-keys

HIC

[SunilChauhan]

yes in some cases

one syn key may not impact data

fo example: when you use intervalmatch it left with one syn key

hope this helps

[its_anandrjs]

Absolutely No it will help you for your data model but if there are many synthetic keys then there are many synthetic tables then they will occur problem for your data model. When you reload your data model it will slow because the synthetic table is maintain in the data model also if your data model is not perfect like primary key or foreign key concept of the SQL is not meet the perfect result may also effect your data model.

But synthetic keys help you to identify the key reference among the multiple tables.

Regards

[SunilChauhan]

Hi  Anand,

I agree with your points.

but if see the link below and please t me provide your view on syn key

hope this helps to understand the impact of syn key

http://community.qlik.com/thread/10279?start=30&tstart=0

[SunilChauhan]

one more question if they are not impacting then why qliktech recommend to remove them.

i thing synthetic key may lead to poor data model and poor data model may lead to poor results also.

plase correct me if i am wrong somewhere

[hic]

My view - and my answer to John's thread - can be read here:

http://community.qlik.com/blogs/qlikviewdesignblog/2013/04/16/synthetic-keys

John writes that he believes that the problems are caused by bad data models, not by the synthetic keys in themselves, and concludes by "Who has an actual problem CAUSED by synthetic keys, rather than by an underlying data model problem?"

I absolutely agree with John. And I have been working in the core team of the QlikView development for twenty years... I was there when we (QlikTech R&D) introduced the synthetic keys.

The bottom line is that Synthetic keys are good! But if you have a bad data model where QlikView creates synthetic keys based on other synthetic keys, then QlikView eats memory and your application is unusable. And the solution is to fix your data model.

HIC

[its_anandrjs]

From my point of view, Yes it is right that to remove them and i also agree with Henric. If there is lot of synthetic keys and more synthetic tables are created then need to remove them. Because it affects the data model also if data model is not proper then my cause bad effect to the data model. Also if there is less tables then synthetic keys are much helper and easier to find the relationship between table and keys for large data model where many tables and huge data will be give performance issue to you.

And you could find that there is more keys are present which has relation between the tables by synthetic tables they will help you for this.

They are helpful to find if  two or many same keys are present in the tables then they created the synthetic keys, then bu synthetic keys or table we identified that the keys are present in more then tables.

See more information about synthetic keys on community.

Thanks & Regards

Anand