It's perfectly valid to mention the higher memory requirements as a considerable drawback.
But this is far from the possible performance that Logstash offers, so the comparison is at least misleading.
Also, there is no information about the memory usage of the other contestants, which probably don't have memory limits if they don't run on the JVM.
I say all this as someone who stopped using Logstash, in part due to the high requirements - but benchmarks should strive for a fair comparison.
