SAC - How to reuse and schedule refresh on existing stories (created via drag and drop)

[VirajSaha]

Hello Experts,

We have a quite a few stories created by our FP&A user group, via drag and drop of local files (excel and csv). We have a requirement to reuse these data and if possible connect BW system for schedule refresh.

We understand that below are the limitations, which kind of sad as some of these stories are well designed on already cleansed data -

• We can not schedule refresh. So we are looking for other options to convert them into models.
• Only way to facilitate schedule refresh from our BW4 environment, if we build the entire setup from scratch.
• We can not convert them into models based setup. We can not use stories as data source to a model.
• We can not link data (residing in these stories) to other models (other way around is possible).

I already checked couple of Q&A threads and everywhere it is mentioned we can not schedule and we can not convert.
Wanted to check if there is any workaround to reuse data from these wonderful stories.

Thanks

[DebjitSingha · Accepted solution]

Hello viraj_saha,

Couple of assumptions in your question are not entirely correct. For example

• We can not schedule refresh. So we are looking for other options to convert them into models. You can schedule refresh if you publish to a model. There is a trick!
• Only way to facilitate schedule refresh from our BW4 environment, if we build the entire setup from scratch. You do not have to build the model from scratch, you can convert it from the same dataset. With some limitations and rework. Explained below.
• We can not convert them into models based setup. We can not use stories as data source to a model. Again you can convert dataset into model. Though story components need to remapped manually.
• We can not link data (residing in these stories) to other models (other way around is possible). You can convert embedded dataset (cared by system) to public dataset. Once converted you will be able to link this data to other stories and models.

Here is how it works. When we create a story based on drag and drop, system end up creating a "dataset" (also known as embedded dataset). This will remain part of story and not available for outside consumption. Embedded dataset allow linking, refreshing, changing data source etc.

What it can not do -

1. Allow other stories (and models) to link (join) with this data set. This is due to the fact data is part of story and embedded dataset is not available independently.
2. Embedded dataset do not allow scheduled refresh from a data source. In fact scheduled refresh is not supported on dataset altogether. One can only refresh data manually.

Lets go through each limitations. Linked analysis and combine data (model level) is only possible if data reside in models or "dataset". Embedded dataset let you convert to "Public Dataset"

Open story > Data> Grid view> Convert to public dataset.

Post conversion, story and dataset are decoupled. Dataset becomes available for consumption in other models / stories (join). Good thing is, newly created public dataset remain linked to existing story and any change / update in dataset impacts the existing story. No remapping of stories required.

Now scheduled refresh will only be possible if we convert embedded dataset to a model. In general this conversion is not supported.

Here is the trick: Story > Data > Grid View> Click on Save > "Open with basic data preparation" > Publish to model.

Now you have a converted model created from a copy of the embedded dataset. Here you can link your BW system and schedule a refresh job.

Limitations: Once you switch to basic data preparation, data gets reset to initial state. All transformation added later are ignored. This means you have to recreate transformation before or after publishing to model. Unlike Public dataset conversion, existing story do not link to new model. This is a manual step. Difference in behavior, partly due to fact "Public Dataset" is a conversion process and "Publish Model" is a copy activity.

Based on your requirement you can decide if you need to convert to public dataset or to a model.

Hope this helps. Let us know if you have any query around this topic.

Viewers/ readers - If you find above information helpful, feel free to up-vote (arrows on left side).

Thanks,

Debjit

[VirajSaha · Accepted solution]

Thanks for your suggestion.

For now we are going to convert most of our stories into public dataset. Select few will consider model conversion. I will vote for the enhancement as well. Will be useful in our scenario.