data.bash

# No caching at all
~/c/ores-stress $ python3.4 /tmp/test.py 'data/tools-static.wmflabs.org/lolrrit-wm/1/*.out'
inflights
mean 1.756802902571764
median 1
pstdev 2.4299546779621792
latency
mean 1724.702838648917
median 838.930203
pstdev 2159.033094872801
# nginx caching, no delay / priming
~/c/ores-stress $ python3.4 /tmp/test.py 'data/tools-static.wmflabs.org/lolrrit-wm/2/*.out'
inflights
mean 2.026774252704933
median 1
pstdev 2.230940052757951
latency
mean 1701.3539797003482
median 963.50666
pstdev 1938.0075171622725
# nginx caching, delay
~/c/ores-stress $ python3.4 /tmp/test.py 'data/tools-static.wmflabs.org/lolrrit-wm/3/*.out'
inflights
mean 3.918343653250774
median 2.0
pstdev 4.978816212595768
latency
mean 2417.0321543610685
median 885.054787
pstdev 3648.74529846867
# nginx caching, priming + delay
~/c/ores-stress $ python3.4 /tmp/test.py 'data/tools-static.wmflabs.org/lolrrit-wm/4/*.out'
inflights
mean 2.1061302681992338
median 1.0
pstdev 2.421693299147825
latency
mean 2022.3888741293106
median 1045.764082
pstdev 2503.281584788331

test.py

import sys
import statistics
import glob

import csv

files = glob.glob(sys.argv[1])
data = []

for file in files:
    with open(file) as f:
        reader = csv.reader(f)
        for row in reader:
            data.append((int(row[0]), float(row[1])))

inflights = [r[0] for r in data]
latency = [r[1] for r in data]


def duh(l):
    print('mean ' + str(statistics.mean(l)))
    print('median ' + str(statistics.median(l)))
    print('pstdev ' + str(statistics.pstdev(l)))

print('inflights')
duh(inflights)
print('latency')
duh(latency)