[Python] boto3 timestream queries very slow

import boto3

device_list = ['ARMS-GFY-D5', 'ARMS-GFY-D3']

client = boto3.client('timestream-query')

for device_id in device_list:
query = f"""
SELECT time, charge, current, voltage, temperature
FROM "telemetryDatabase"."telemetryTable"
WHERE device_id = '{device_id}'
AND time > ago(24h)
ORDER BY time
"""

response = client.query(QueryString=query)

• The query above entered into the AWS timestream console query editor (with device ID substituted) consistently completes in under 1 second.
• The first device_id query executed via this python script from my dev laptop consistently takes >15s, often as much as 30-60s.
• The second device_id query executed via this python script consistently returns in <1s

In all cases the output is small, only about 12kB (288 rows). Reducing the time range of the query has no impact on query time.

My laptop has a gigabit internet connection and no obvious network throughput issues to AWS.

The table schema:

Why is boto3 so much slower than AWS console, seemingly only until the client is "warm"? How can I avoid this behavior?

Continue reading...