Slow process of broker with less load

Hi Team,
We are not using kubernetes approach to run Zeebe components due to some limitations, we have 8 cores CPU and 16gb memory infra. And we are running 2 brokers, 1 gateway, 1 elastic search , operate and tasklist.
We observed it is taking more time broker to do the processing. We created 20 parallel threads and created process instance which has 5 tasks still observed latency is very high , which was more than 1 sec to create process instance and activate the next service task.

Can we get some help in what infra we need to keep on both zeebe components side and client side to achieve 200process instance throughput per second. Throughput I mean, creation to completion for process instance should be less than 200ms for a process instance which is having 3 service tasks. (Service task is light weight which will not take more 30ms to execute).

Let me know if any other details are required.