QLIK COMPOSE IS ETL OR ELT

[deepaksahirwar]

Dear Community Users,

Would like to discuss topic on the group whether Qlik Compose be called as ETL OR ELT operation.

As per Qlik terminology which one best suites.

Thanks in advance.

Best Regards,

Deepak 

[sureshkumar]

Hello @deepaksahirwar 

Qlik Compose is a ETL Operation.

Regards,

Suresh

[deepaksahirwar]

Dear @sureshkumar ,

Thanks for responding.

But you know some of the business called it as 'ELT' tool.

So can compose be already used in best as 'ELT'

Best Regards,

Deepak

[TimGarrod]

Hi, Qlik Compose processes data using an ELT paradigm.   It generates set-based push-down SQL to execute against your data warehouse (Snowflake / BigQuery/ SQL Server etc.).

Qlik Replicate and its CDC capability is leveraged to perform the E-L in the Qlik Data Integration architecture and then Qlik Compose provides the 'T' with its automated data warehouse capabilities. 

Qlik Cloud Data Integration operates in a similar fashion, however the CDC / data loading and transformations are more tightly coupled in an end-to-end data pipeline design scenario. 

[deepaksahirwar]

Hi @TimGarrod ,

You mean Compose is ETL only.

My other query based on your response.

Can qlik replicate be also called as 'ETL' , because it has some built in transformations available which can be utilised as per demand.

Thank you,

Deepak 

[TimGarrod · Accepted solution]

Compose is ELT only. It does not process via an ETL paradigm.

ETL requires an engine (typically mid tier) that extracts data from sources, transform in the engine and then loads to the target.

Compose is ONLY ELT - using the target data warehouse platform as the engine for data processing (push down sql). No data is ever extracted from a source into Compose for processing However as I said, Compose does not really perform the “E” - extract from a disparate source system. All its “source” data needs to exist in the target. Which is where Replicate comes in.


Replicate is a replication and CDC technology. I hesitate to say it’s ETL because the “T” in replicate is very light. You cannot join 2 tables together for example which is a fundamental feature of an ETL technology. That maybe semantics but not if you were to evaluate Replicate for ETL use cases.

My 2 cents 🙂