I’m looking for some documentation to perform a Kafka capacity plan and hardware sizing.
I have a lot of doubts about the memory and CPU sizing. About the memory I know to calculate the total cluster memory requirement M = w*b, where w is the write_throughput and b is the seconds of buffer. The doubt is: can I divide M by x, where x is the memory of a single node, to calculate the number of node with x memory i need in my cluster?
For example, if I have: W=200 Mb/sec, b=30 sec => W*b=5.86 Gb/sec.
If I divide 5.86/2 = 2.9, it means that i need 3 server of at least 2 Gb of memory in my cluster?
Thanks!