quoracle/examples/case_study.py
2021-02-07 12:19:08 -08:00

131 lines
4.2 KiB
Python

"""
This script contains the code used in the case study of our paper
(https://mwhittaker.github.io/publications/quoracle.pdf).
"""
# See https://stackoverflow.com/a/19521297/3187068
import matplotlib
matplotlib.use('pdf')
font = {'size': 8}
matplotlib.rc('font', **font)
from quoracle import *
import datetime
import matplotlib.pyplot as plt
def main() -> None:
def seconds(x: int) -> datetime.timedelta:
return datetime.timedelta(seconds=x)
a = Node('a', write_capacity=2000, read_capacity=4000, latency=seconds(1))
b = Node('b', write_capacity=1000, read_capacity=2000, latency=seconds(1))
c = Node('c', write_capacity=2000, read_capacity=4000, latency=seconds(3))
d = Node('d', write_capacity=1000, read_capacity=2000, latency=seconds(4))
e = Node('e', write_capacity=2000, read_capacity=4000, latency=seconds(5))
fr = {
1.00: 0.,
0.90: 10.,
0.80: 20.,
0.70: 100.,
0.60: 100.,
0.50: 100.,
0.40: 60.,
0.30: 30.,
0.20: 30.,
0.10: 20.,
0.00: 0.,
}
maj = QuorumSystem(reads=majority([a, b, c, d, e]))
grid = QuorumSystem(reads=a*b + c*d*e)
paths = QuorumSystem(reads=a*b + a*c*e + d*e + d*c*b)
print('0-resilient Capacities')
print(maj.uniform_strategy().capacity(read_fraction=fr))
print(maj.capacity(read_fraction=fr))
print(grid.capacity(read_fraction=fr))
print(paths.capacity(read_fraction=fr))
print()
print('0-resilient Searched')
start = datetime.datetime.now()
opt, _ = search(nodes=[a, b, c, d, e],
resilience=1,
read_fraction=fr)
stop = datetime.datetime.now()
print((stop - start))
sigma = opt.strategy(read_fraction=fr)
print(opt)
print(sigma)
print(sigma.capacity(read_fraction=fr))
print()
for (sigma, name, filename, size) in [
(maj.uniform_strategy(),
'Majority Quorum System',
'majority_uniform',
(3.25, 2)),
(sigma,
'Searched Quorum System',
'searched',
(3.25, 1.75)),
]:
fig, ax = plt.subplots(figsize=size)
plot_node_throughput_on(
ax,
sigma,
nodes = [a, b, c, d, e],
read_fraction=0.5,
draw_node_capacities=False,
)
ax.set_xlabel('Node')
ax.set_ylabel('Throughput\n(commands per second)')
fig.tight_layout()
fig.savefig(f'{filename}_throughputs.pdf')
print('1-resilient Capacities')
print(maj)
print(maj.capacity(read_fraction=fr, f=1))
print(grid.capacity(read_fraction=fr, f=1))
print(paths.capacity(read_fraction=fr, f=1))
print()
print('1-resilient Searched')
start = datetime.datetime.now()
opt, _ = search(nodes=[a, b, c, d, e], resilience=1, read_fraction=fr, f=1)
stop = datetime.datetime.now()
print(stop - start)
sigma = opt.strategy(read_fraction=fr, f=1)
print(opt)
print(sigma)
print(sigma.capacity(read_fraction=fr))
print()
print('Latency Optimal Capacities and Latencies')
print(maj.uniform_strategy().capacity(read_fraction=fr),
maj.uniform_strategy().latency(read_fraction=fr))
print(maj.capacity(read_fraction=fr, optimize='latency', load_limit=1/2000),
maj.latency(read_fraction=fr, optimize='latency', load_limit=1/2000))
print(grid.capacity(read_fraction=fr, optimize='latency', load_limit=1/2000),
grid.latency(read_fraction=fr, optimize='latency', load_limit=1/2000))
print(paths.capacity(read_fraction=fr, optimize='latency', load_limit=1/2000),
paths.latency(read_fraction=fr, optimize='latency', load_limit=1/2000))
print()
print('Latency Optimal Searched')
start = datetime.datetime.now()
opt, _ = search(nodes=[a, b, c, d, e], resilience=1, read_fraction=fr,
optimize='latency', load_limit=1/2000)
stop = datetime.datetime.now()
print(stop - start)
sigma = opt.strategy(read_fraction=fr, optimize='latency', load_limit=1/2000)
print(opt)
print(sigma)
print(sigma.capacity(read_fraction=fr))
print(sigma.latency(read_fraction=fr))
print()
if __name__ == '__main__':
main()