Forum Discussion
Understanding the bucket size setting for the incremental refresh in dataflow gen2
- Anonymous1 year ago
Hi pmscorca ,
Thanks for sharing your incremental refresh settings. Let's break down your points to clarify the concepts further.
- Monthly Bucket and Date Column to Filter By: You're correct that a monthly bucket means the data will be partitioned by month based on the "orderdate" column. If you have orders only in August, the partition will be for August. However, if you have orders in August, September, and October, there will be three partitions, one for each month. The "Extract data from the past" option is indeed relevant for the initial load, ensuring that historical data is included based on the specified range.
- Extract Data from the Past: This setting is primarily for the initial load. For subsequent loads, the data will be loaded based on the current month and moving forward. For example, in October, it will load October's orders, in November, it will load November's orders, and so on. The order date will determine which monthly bucket the data falls into, such as an order with the date 10/20/2024 falling into the October bucket.
- Maximum Value in Modified Date Column: This setting ensures that if any data within a monthly bucket changes, the entire bucket is refreshed. So, if an order within a monthly bucket is modified, the entire bucket for that month will be refreshed, potentially replacing unmodified orders if any changes are detected within that bucket.
Best Regards,
Adamk KongIf this post helps, then please consider Accept it as the solution to help the other members find it more quickly
Hi, thanks for your reply. I've opened this post after reading the related documentation and doing some proofs.
I'd like to understand better this feature.
I need to import some excel files as a source in order to write the related data in a Fabric warehouse.
As a first load, I've these data:
two orders at August 10th.
Inside the dataflow gen 2 I have these settings:
- Date to filter by = orderdate,
- Extract data from the past = 2 months,
- Bucket size = month,
- Only extract new data when the maximum value in this column changes = modifieddate.
Which are the splitted bucket ranges respect to the order date, in terms of start date and end date?
Thanks
Excel files are not a foldable source. Incremental refresh should only be used for large data, buckets should have at least 2 Million rows for this to make sense.
- pmscorca1 year ago
Post Prodigy
Hi, I'm trying with few rows to understand better the mechanism and however I've no issues using excel files.