multithread scan table spend so much time

[gangchzu]

my code is below:
from concurrent.futures import thread
from typing import List
from pynamodb.attributes import UnicodeAttribute, NumberAttribute
from pynamodb.models import Model

import time
import threading

class WorkBaseModel(Model):
instance_id = UnicodeAttribute(default="", hash_key=True)
worker_id = UnicodeAttribute(default="", range_key=True)
version = UnicodeAttribute(default="0")
region_name = UnicodeAttribute(default="")
instance_state = UnicodeAttribute(default="")
work_state = NumberAttribute(default=0)
last_report_time = NumberAttribute(default=0)
create_time = NumberAttribute(default=0)
protected = NumberAttribute(default=0)
instance_type = UnicodeAttribute(default="")

class MetaBase:
aws_access_key_id = ""
aws_secret_access_key = ""
max_retry_attempts = 6
max_pool_connections = 40

class ConferenceAssistantInstance(Model):
instance_id = UnicodeAttribute(default="", hash_key=True)
instance_state = UnicodeAttribute(default="")
create_time = NumberAttribute()
protected = NumberAttribute(default=0)
region_name = UnicodeAttribute(default="")

class Meta(MetaBase):
    table_name = "conference_assistant_instance"
    region = "ap-northeast-1"

if name == "main":

def run(i):
    now = time.time()
    all_instance_list: List[ConferenceAssistantInstance] = list(
        ConferenceAssistantInstance.scan()
    )
    print(len(all_instance_list))
    print(f"user time: {i}", time.time()-now)

threads = []
for i in range(40):
    t = threading.Thread(target=run, args=(i,))
    threads.append(t)
    t.start()
    
for thread in threads:
    thread.join()
    
print('over')

I have 100 items in table conference_assistant_instance, when I run this script, spend time is about 3 seconds, but when I replace scan by boto3, it is just 100 milliseconds

[jbsilva]

Could you test with version 5.5.1?
Maybe this performance issue is related to my issue #1253