Multiple SSLSocket I/O in Threads: the slowest connection universally sets the pace for all connections

I’m curious… how are you measuring throughput?

The numbers I’m mentioning are per download. When I’m using Firefox, it’s what it says on the active downloads status. I consider it valid if it has somewhat stabilized around a certain level after a few refreshes.
In some other cases, I’m using another xojo-made tool that downloads via the URLConnection. There, I’m getting the duration of the download in milliseconds.

Are my figures dissimilar to the ones you’re seeing?

No. I just wanted to make sure you weren’t doing anything that would have skewed the results. Browsers will sometimes report speeds based on uncompressed sizes even if the content is compressed and the download speeds will be falsely inflated. So I was just checking.