Which is faster between join and keep

[debsahoo]

>For load script  which is faster between Join and keep? why?

>For dashboard which is faster? why?

Thanks in advance.

[Anil_Babu_Samineni]

In theoretical Join and Keep are two different concepts. Not sure, What you are expecting?

Which is faster, Means what my dear?

[pradosh_thakur]

Hi deb

lets assume you have a large data and you are asking this question referring to that

As per my understanding

1:if you are using keep it will be less costly and less time taking during the script execution

2: By join you may get the data updated faster when any filter is applied on the front end compared to keep.

3: There is no outer keep so you cant compare that.

Hopefully it will be helpful to you.Experts can correct me if i am wrong here.

regards

Pradosh

[martinpohl]

with join you can add datas to a present table.

with keep you can load datas and reduce datas in another table.

at dashboard there is no difference because join and keep are scripte statements

remember:

Qlik ist different than other BO tools.

The possibility, to keep datas in different tables, linked by fields to create charts and tables makes it easier to create a data model without thinking in cubes or wide, large tables.

Keep datas in tables as you can and feel free to create dashboards on it.

[marcus_sommer]

Quite often is mapping a lot faster than joining - even if it's need multiple and/or combined mappings it will run quite fast: Mapping as an Alternative to Joining.

- Marcus

[debsahoo]

Means which performance is faster for large data. If there are two table having some millions of record, in this case which is good between keep and join to load data faster.

I am new in qlik and try to learn more. Plz help me to understand join and keep.

Mostly in which scenario we will prefer  for Keep?

[marcus_sommer]

Your question is a bit too general to give a concrete answer. It will depend on various things mostly how the datamodel should look like which ways of building and associate the tables is most suitable.

Personally I would try to avoid such large joinings and using other ways like below mentioned with a mapping-approach.

- Marcus

[debsahoo]

If both table have more than 10 fields only a KEY field is common of them. That case mapping Possible.?

[marcus_sommer]

In general, yes it's possible. You could apply multiple mappings and also you could concat several fields like

Map:

mapping load Key, F1 & '|' & F2 & '|' & F3 from Source;

and then using subfield(applymap('Map', Key, '#NV'), '|', 1) to grab for example the first field of the concatenated lookup-value.

Further mapping has a very important advantage against joining-approaches because you don't need to worry about duplicating or removing records by choosing the wrong joining-type and/or by issues with the data-quality.

Beside this you should consider if you really need this kind of merged table to get your datamodel to work.

- Marcus

[debsahoo]

Yes I am able to do mapping, thanks for reply but I want to know more about join and keep. And in which scenario we will go for keep instead of join.