How can I move my source tables to another service?

[Anonymous]

Hello

I have some tables in Hadoop and I want to move them to Azure.

These tables are input as sources in PowerBI.

How can I make the transition without having to rebuild the dashboards?

Thanks!

[edhans]

I suggest you copy the files to Azure and leave the on Hadoop during the transition.

1. Create connections to the files/data sources however you would on Azure. You didn't say how you were connecting. Azure Data Lake, Storage Blob, SQL, etc.
2. do the transformations in Azure to get it to the first point in the Hadoop transformations. usually this will be the first and possibly second step in the Advanced query editor.
3. Take those 1/2 steps and replace the first few (1 to 3 stesp) of the Hadoop transformation. Make sure the last step you copy from your Azure transformations has a comma at the end.
4. You can delete the query used to generate the Azure transformations. 

You can follow more detailed steps in this article, or you can post your Hadoop M code here and your Azure M code here and we can help walk you through specifics.

See this article if you need help using this M code in your model.

[Anonymous]

@edhans thanks, my code is below. Can I just replace the first 4 lines until the #"Renamed Columns"...?

    Source = Odbc.DataSource("dsn=f", [HierarchicalNavigation=true]),
    IMPALA_Database = Source{[Name="IMPALA",Kind="Database"]}[Data],
    workspace_Schema = IMPALA_Database{[Name="workspace",Kind="Schema"]}[Data],
    table1_Table = workspace_Schema{[Name="table1",Kind="Table"]}[Data],
    #"Renamed Columns" = Table.RenameColumns(........
    #"Trimmed Text" = Table.TransformColumns(#"Renamed Columns",{...
    #"Remove e" = Table.ReplaceValue(...
    #"Remove r" from ID" = Table.ReplaceValue(...
    #"Replaced Value" = Table.ReplaceValue(...
in
    #"Replaced Value"

[edhans]

Exactly @Anonymous 

Build the connection using the new method and get that new connection to the "Renamed Columns" step. As long as both tables have the same columns to that point, just swap out the M code.

The only caveat is check on the data types. Different sources can return different data types. SQL returns good data types you can rely on, but Excel does not, so you have to change. Just make sure that if your columns had good data types on the initial connection, you validate them when you swap out the connection. Dealing with databases it should not be an issue, but double check so you don't hit any calculation errors, refresh errors, or issues where something silently breaks.

For example, a Date field that converts to text will cause all date time intelligence to stop working, but it just provides wrong answers, not errors.

[VijayP]

@Anonymous 

https://docs.microsoft.com/en-us/azure/hdinsight/hadoop/apache-hadoop-on-premises-migration-best-practices-data-migration

This article may help you!