prices.py

from collections import defaultdict
from datetime import datetime, timedelta
from decimal import Decimal
from pprint import pformat, pprint
import json

import boto3
import requests

with requests.get("http://www.ec2instances.info/instances.json") as response:
    vCPUs = {
        instance["instance_type"]: instance["vCPU"]
        for instance in response.json()
    }


clients = {
    region["RegionName"]: boto3.client("ec2", region_name=region["RegionName"])
    for region in boto3.client(
        "ec2", region_name="us-east-1",
    ).describe_regions()["Regions"]
    if region["RegionName"].startswith(("us-east", "us-west"))
}
availability_zones = {
    region: {
        zone["ZoneName"]
        for zone in client.describe_availability_zones()["AvailabilityZones"]
    } for region, client in clients.iteritems()
}


price_per_vCPU = {}

for instance, vCPU in vCPUs.iteritems():
    price_per_vCPU[instance] = {}
    for region, client in clients.iteritems():
        price_per_vCPU[instance][region] = region_prices = {}
        for zone in availability_zones[region]:
            history = client.describe_spot_price_history(
                        InstanceTypes=[instance],
                        AvailabilityZone=zone,
                        ProductDescriptions=["Linux/UNIX (Amazon VPC)"],
                        MaxResults=1,
                        StartTime=datetime.utcnow() - timedelta(hours=1),
            )["SpotPriceHistory"]
            if history:
                price, = history
                region_prices[zone] = Decimal(price["SpotPrice"]) / vCPU

        top_3 = sorted(
            (price, zone) for zone, price in region_prices.iteritems()
        )[:3]

        if not top_3:
            continue