There's a balance between compression ratio and CPU utilisation. Zstd is several times faster at compressing than gzip to produce a file of similar size (it's good, give it a try if you haven't already). I guess he means by using zstd they were able to crank up the compression ratio and maintain the same CPU usage maybe?
I don't think AWS routinely compress customer data that I've noticed. I guess he must mean for their internal products that use S3 perhaps?
I don't think AWS routinely compress customer data that I've noticed. I guess he must mean for their internal products that use S3 perhaps?