Thomas, I don't know the figures but there must be minimal hundreds of AWS Hana Developer instances based on the recommended m2.2xlarge sizing - if the problem is as fundamental, core and generic as Wolfgang alludes to then you would agree that all these hundreds of instances, where "where memory is very tight", are affected, yes?
If yes, then where are the hundreds of SP6 developers screaming about this? That's why it doesn't add up for me at least, yet happy to be enlightened.
Earlier in this thread I highlighted sluggish performance with an m2.4xlarge instance - that's double the memory capacity of recommended standard, with 68Gb. I will continue to develop for the next week with an m2.4xlarge instance, adjusted with higher memory allocation to the statistics process, as per Wolfgang's workaround, and see how it goes.
Perhaps a general consideration for now to "up" the default sizing to m2.4xlarge along with Wolfgang's workaround?
Thanks.