Confused about the statement that QPS of 23K is not easily feasible


#1

Please help me understand why “QPS of 23000 is not easily feasible”. As far as I am concerned memcached can easily sustain 100K QPS and more (e.g. https://github.com/memcached/memcached/wiki/Performance#handles-extremely-high-load )


#2

My understanding is that in the table of latency numbers they provided, you can’t query 1MB in each query because that costs 250us + overhead.