Sounds like they were using the Range header on large files. I have made systems in the past using exactly this pattern(without the intentionally dropped reques

Sounds like they were using the Range header on large files. I have made systems in the past using exactly this pattern(without the intentionally dropped requests). I hope this doesn't result in any significant changes as I really liked using this pattern for sequential data processing of potentially large blobs.