Threads won't get you far...

moving_threads.py

from multiprocessing import Pool
import boto3


def move_to_glacier(f):

  sess = boto3.session.Session(region_name='us-east-1')
  s3res = sess.resource('s3')

  copy_source = {
    'Bucket': f[0],
    'Key': f[1]
  }
  s3res.meta.client.copy(
    CopySource=copy_source,
    Bucket='my-destination-bucket',
    Key=f[1],
    ExtraArgs={'StorageClass': 'GLACIER'}
  )
  
  ...
  
gazillion_files = [('my-source-bucket', 'file-000000000001'), ...]

p = Pool(8)
p.map(move_to_glacier, gazillion_files)