cost_estimator.py

from pymongo import MongoClient
import math
import datetime
from collections import namedtuple
import calendar
db = MongoClient()['mci']

NUM_THRESHOLD = 30
start = datetime.datetime(2014, 4, 1, 0, 0, 0, 0)
end = datetime.datetime(2014, 5, 1, 0, 0, 0, 0)

size_pricing = {
        "m3.2xlarge" : 1.064,
        "m3.xlarge" : 0.532,
        "m3.large" : 0.266,
        }

TERMINATED = -1
STARTED = 1

hosts = list(db.hosts.find({"started_by":"mci","creation_time":{"$gte":start, "$lt":end}, "distro_id":{"$in":["windows-64-compile", "windows-64-test"]}}, sort=[("creation_time", 1)]))

HostEvent = namedtuple('HostEvent', ['type', 'time', 'size', 'cost'])

all_host_events = []

for host in hosts:
    hourly_cost = size_pricing[host['instance_type']]
    billed_hours = math.ceil((host['termination_time'] - host['creation_time']).total_seconds() / 3600.)
    cost = billed_hours * hourly_cost
    started = HostEvent(STARTED, host['creation_time'], host['instance_type'], cost)
    terminated = HostEvent(TERMINATED, host['termination_time'], host['instance_type'], 0)
    all_host_events.append(started)
    all_host_events.append(terminated)
    
all_host_events.sort(lambda h1,h2: int(math.ceil((h1.time - h2.time).total_seconds())))


cur_max = 0
total_hosts = 0
total_cost_estimate = 0
discounted_cost_estimate = 0
for h in all_host_events:
    total_cost_estimate += h.cost
    total_hosts += h.type
    if total_hosts > cur_max:
        cur_max = total_hosts
    if total_hosts > NUM_THRESHOLD:
        discounted_cost_estimate += h.cost

print total_cost_estimate
print discounted_cost_estimate