Spotfire Load on Demand (multiple data sources)-Selecting order

[Krista Dobley]

I have a dashboard which has many "load on demand" data sources against large data warehouses, with a "Load" action control button to allow the user to process a load data from an input field (multiple lines), which now is taking ~30 minutes to return data for them all. Are there lines I can add to a python script to select the order which the load on demand will process the load against my data sources (I'll load the quickest ones first), or even for a user to be able to select which data sources they want to reload against the data they post in the input box? Thank you!

[David Boot-Olazabal]

Hi Krista,

Not sure how your script looks like, but I have found this one that defines the tables to be loaded:|
 

from Spotfire.Dxp.Data import DataManager,DataTable
from Spotfire.Dxp.Application.Scripting import ScriptDefinition
from System.Collections.Generic import Dictionary,List
from System.Collections import ArrayList
from Spotfire.Dxp.Framework.ApplicationModel import NotificationService
import clr

table=List[DataTable]()
table.Add(Document.Data.Tables["test"])
Document.Data.Tables.Refresh(table)

When you copy the 'table.add' line to provide for all the data on-demand tables, it may just follow that order to kick off the loading of data for these tables. I am assuming that you have deselected the boxes "Load automatically" and "Allow caching" for all your on-demand tables.

There might be possible a workaround, if you want to give users more freedom to select the tables that should be updated. In this case, you have to create a table holding all the on-demand table names. By marking specific table names, the iron python script should then use these values to kick off the on-demand load for the selected tables. As I have not tested this myself, I am not sure if and how you would implement that in the iron python script. And I do not know if you can build something into the code, that presets the loading of a specified set of on-demand tables, based on the user that is opening the analysis.

On a different note, permitting this is possible in your case, I would encourage you to have a look at the long loading on-demand queries. As you mentioned, it takes quite a while to load all the on-demand tables against a large data warehouses, which kind of defeats the benefits of on-demand. Since data warehouses are not being updated as frequently as a transactional database, would it be an option to use scheduled updates or automation services jobs of the bigger queries, which are then used in the analysis?

Kind regards,

David