Benchmarks<\/h2>\n\n\n\n
For comparison, I imported 10000 contacts with the fields „First Name“, „Last Name“, „Job Title“, „Email“, „Business Phone“, „Gender“ and „Birthday“. The dataset used was a sample people dataset from Datablist<\/a>. I ran each import method three times to calculate an average time:<\/p>\n\n\n\n Adding lookup and choice fields increased the import time by about 30% for all methods.<\/p>\n\n\n\n When using Bulk Messages, there’s four primary steps you’ll need to execute, which can be wrapped in a Power App on demand trigger:<\/p>\n\n\n\n\n
![\"\"](\"https:\/\/niklas-boesch.de\/wp-content\/uploads\/2024\/09\/chart.png\")
<\/p>\n\n\n\nUsing Bulk Messages<\/h2>\n\n\n\n
\n
In our example, this is done using the „List rows present in a table“ action with the Excel connector, using an xlsx file saved in OneDrive, but it can be any kind of action which results in you having your data in the flow.![\"\"](\"https:\/\/niklas-boesch.de\/wp-content\/uploads\/2024\/09\/List-Rows-from-Excel.png\")
<\/li>\n\n\n\n
If the data is already available as an array and you only need to map to the right column names, this is a simple „Select“ step. For each field you want to fill in your record, you’ll need to map the logical name of the field to your source data. However, there are some special cases:
\n\n
To import dates, use the formatDateTime<\/code> expression like this: formatDateTime(item()?['Date of birth'], 'yyyy-MM-dd')<\/code><\/li>\n\n\n\n
You’ll need to set choice fields using the numerical Option Set Value (In our example I’m just setting gendercode to 1, which is the code for male). These values can be seen through the maker portal, or with the great Level up for Dynamics 365\/Power Apps<\/a> browser add in.<\/li>\n\n\n\n
To connect to other records, your formatted data needs to use OData Binding notation. For a detailed overview of OData Binding notation, check out Syed Sajid Hussain’s blog post<\/a>.
In our example, I’m using a fixed GUID, but it could also be a value from the source data or an alternate key.<\/li>\n\n\n\n
For every record which is to be imported, you’ll need to add the „@odata.type“ property, with a value of „Microsoft.Dynamics.CRM.[entitylogicalname]“. In our example, this is going to be „Microsoft.Dynamics.CRM.contact“. As you can see in the screenshot below, to add the property with the select step, we need to repeat the „@“, so Power Automate does not confuse the string with an attempt to create an expression.![\"\"](\"https:\/\/niklas-boesch.de\/wp-content\/uploads\/2024\/09\/Format-Source-Data-1.png\")
<\/li>\n<\/ul>\n<\/li>\n\n\n\n
To avoid message size and time limits<\/a> large inputs need to be chunked and imported separately.
There’s a trade-off between large chunks leading to shorter import times, but increasing the chance that your import is going to fail because of those limits. Microsoft recommends 100-1000 records per message, depending on size of records and any plugins which may run during the import.
To find the sweet spot, I recommend setting the chunk length in a separate variable and experimenting with the most complicated subset of your import data. Don’t forget to have a margin of error, as the API and your environment does not have a fixed throughput.![\"\"](\"https:\/\/niklas-boesch.de\/wp-content\/uploads\/2024\/09\/Target-Chunk-Array.png\")
<\/li>\n\n\n\n
After chunking, only the import is left. For that, an apply to each loop over our chunked data is used, containing only the „Invoke an HTTP request“ with the preauthorized Entra ID connector.
The request can be set as seen below, please note that the entity set name needs to be a part of the url of the request.
The loop itself can be run in parallel as well to further speed up the process. I’ve used a concurrency level of 4 to keep to the Service protection limit<\/a> of 20 minutes execution time per 5 minute sliding window.![\"\"](\"https:\/\/niklas-boesch.de\/wp-content\/uploads\/2024\/09\/Import-Data.png\")
<\/li>\n<\/ul>\n\n\n\n