stats.py

#!/usr/bin/env python

import os
from datetime import datetime

total_objects = 4300000

elapsed = 0
last = None
count = 0

# figure out the total elapsed time between requests and the total number
# of requests in order to calculate the average requests / second

for line in open('archive-fedora.log'):

    t = datetime.strptime(line.split(' - ')[0], "%Y-%m-%d %H:%M:%S")
    if last:
        elapsed += (t - last).total_seconds()
    count += 1
    last = t

# count the number of files and their total size to calculate average file size

file_count = 0
file_size_bytes = 0.0
for root, dirs, files in os.walk('archive'):
    for f in files:
        file_count += 1
        file_size_bytes += os.path.getsize(os.path.join(root, f))

reqs_per_sec = count / float(elapsed)
estimated_time_days = (total_objects / reqs_per_sec) / 60 / 60 / 24

avg_file_size_bytes = file_size_bytes / float(file_count)
estimated_size_gb = (total_objects * avg_file_size_bytes)  / 1024 / 1024 / 1024.0

print ""
print "avg req/sec: {}".format(reqs_per_sec)
print "esimated total time: {} days".format(estimated_time_days)
print ""
print "avg bytes/file: {}".format(avg_file_size_bytes)
print "estimated total size: {} GB".format(estimated_size_gb)
print ""