Performance optimization issues during load testing

# Description

I'm experiencing performance issues when conducting load testing on the Conductor service. Here are the details。

# Environment

Conductor version: 3.21.12,
Conductor service node: 4
Operating system: [Debian 5.10.46-4 (2021-08-03) x86_64 GNU/Linux]
Hardware configuration: [CPU 8 cores, 32G RAM, storage 512G]

# Performance Issues Observed
Describe the specific performance problems you're encountering (e.g., high latency, low throughput, timeouts, etc.).
Include any error messages or warnings that appear during testing.

# Load Testing Setup
Tool used for load testing: [JMeter.]
Number of concurrent users/requests: 10/5min
Key metrics observed :

# Configuration

conductor.tasks.http.readTimeout=5000ms
conductor.tasks.http.connectTimeout=3000ms

#Timeout in milliseconds used by
conductor.app.system-task-queue-pop-timeout=0

#The number of threads to use to do background sweep on active workflows.
conductor.app.sweeper-thread-count=2

#The number of threads to be usde in Scheduler used for polling events from multiple event
#queues. By default, a thread count equal to the number of CPU cores is chosen.
conductor.app.event-queue-scheduler-poll-thread-count=2

#The number of messages to be polled from a default event queue in a single operation.
conductor.app.event-queue-poll-count=4

#The max number of the threads to be polled within the threadpool for system task workers.
conductor.app.system-task-max-poll-count=4

#The maximum number of threads to be allocated to the executor service threadpool.
conductor.app.executor-service-max-thread-count=4

# Expected Behavior
JVM config -Xms4096m -Xmx4096m

# Attachments