AzureBlobStorageConnector / Clip LAS file without Download

​@galigis That’s a really good question. Are you doing remotely? Over VPN? That would hurt. :(

I cannot think of a way to limit it by geographical area like one would use an SQL “Where” query to limit database reads. 

This is the same way it would work if it was located on a remote server. FME is going to bring that data local to the machine you are using. Everytime. 

Is there a way to reorganize the data into smaller tiles or chunks? That might be a lot of short-term work but the saving may pay-off.

Hope that helps!

Thank you, ​@crutledge . I successfully accessed the LAS files hosted on the Azure blob using a FeatureReader by specifying the URL path. This method eliminates the need to download the LAS file locally and bypasses the use of the AzureBlobConnector.

I am now exploring options to enhance performance, as reading the LAS blob consumes a significant amount of memory in the workflow. I have implemented a constraint, 'Initiator OGC-Intersects Result,' to read only the portion of the LAS file within the Area of Interest (AOI) and subsequently clip it. However, this process still takes approximately 9 to 10 minutes on average to read the LAS blob. I have tried creating Tiles for the AOI and then used Parallel Processing but didn’t help.

Attached is a screenshot of the complete workflow:

 Any ideas to improve the performance more than welcome! :) 

Hello there ​@galigis, I agree with switching to the FeatureReader. Similarly to the format LAS Reader, you should be able to read with WaaFS + search envelope applied. Are you clipping the input data with a search envelope in the FeatureReader? This option was added sometime in Form 2024, you may need to upgrade to access functionality (FME Downloads Page). Ideally, this would avoid downloading the entire dataset.

Alternatively, have you considered exploring a cloud native options, like COPC? COPC is also built on the LAS standard. The entire idea or drive behind cloud native is to ‘read what you need’, which seems fitting here! Happy to help, Kailin