gistfile1.txt

"""
This script is about telling want you need *today* - not what you want to hear

%(prog)s <number-of-devices-you-have-in-the-real-hardware-you-are-going-to-deploy>

"""
import os
import sys
from math import log, modf, floor, ceil

REPLICAS = 3.0

MAXIMUM_NUMBER_OF_DIRECTORIES_YOU_WANT_IN_TOP_LEVEL_DIR = 7000
MAXIMUM_NUMBER_OF_DIRECTORIES_IN_TOP_LEVEL_DIR = 15000

MIDDLE_OF_THE_ROAD_PARTS_PER_DISK = (
    MAXIMUM_NUMBER_OF_DIRECTORIES_YOU_WANT_IN_TOP_LEVEL_DIR +
    MAXIMUM_NUMBER_OF_DIRECTORIES_IN_TOP_LEVEL_DIR
) / 2

MINIMUM_NUMBER_OF_PARTS_PER_DISK = 100


def main():
    prog = sys.argv.pop(0)
    try:
        device_count = int(sys.argv.pop(0))
    except (IndexError, ValueError):
        sys.exit(__doc__ % {'prog': prog})

    as_big_as_you_really_want_to_go = int(round(log(
        (MAXIMUM_NUMBER_OF_DIRECTORIES_YOU_WANT_IN_TOP_LEVEL_DIR *
         device_count) / REPLICAS, 2)))

    as_big_as_you_can_really_support = int(round(log(
        (MAXIMUM_NUMBER_OF_DIRECTORIES_IN_TOP_LEVEL_DIR *
         device_count) / REPLICAS, 2)))

    number_of_parts = int(round(
        REPLICAS * (2 ** as_big_as_you_really_want_to_go) / device_count
    ))
    part_delta = MIDDLE_OF_THE_ROAD_PARTS_PER_DISK - number_of_parts
    print '@%s' % as_big_as_you_really_want_to_go, 'number of partitions per device', number_of_parts
    number_of_parts_max = int(round(
        REPLICAS * (2 ** as_big_as_you_can_really_support) / device_count
    ))
    part_delta_max = number_of_parts_max - MIDDLE_OF_THE_ROAD_PARTS_PER_DISK
    print '@%s' % as_big_as_you_can_really_support, 'number of partitions per device', number_of_parts_max
    if part_delta < part_delta_max:
        right_answer = as_big_as_you_really_want_to_go
    else:
        right_answer = as_big_as_you_can_really_support

    print right_answer

    max_devices = REPLICAS * (2 ** right_answer) / 100
    thin_nodes = max_devices / 12
    dense_nodes = max_devices / 24
    small_drives = max_devices * REPLICAS
    big_drives = max_devices * 5
    print '%s drives in %s-%s servers storing %s-%s TBs' % tuple(int(x) for x in
                                                              (max_devices,
                                                               dense_nodes,
                                                               thin_nodes,
                                                               small_drives,
                                                               big_drives))


if __name__ == "__main__":
    sys.exit(main())