def parse_results(files: List[str], type: str) -> Dict[str, List[float]]:
data = {
'metal': [],
'kvm': [],
'firecracker': [],
'docker': [],
'gvisor': []
}
for f in files:
selector = get_platform_selector(f)
arr = data[selector]
def parse_line(line: str) -> None:
m = re.search(' +{}, MiB/s: +([0-9\.]+)'.format(type), line)
if m:
mbytes = float(m.group(1))
if mbytes > 0:
arr.append(mbytes)
read_file_per_line(path.join(results_dir, f), parse_line)
assert len(arr) == 10, "Expected to collect 10 results from {} but got {}".format(f, len(arr))
return data
import pandas as pd
dirname = path.dirname(__file__)
results_dir = path.join(dirname, "../tests/network/output")
output_dir = path.join(dirname, "extracted")
bandwidth_file = "bandwidth_results.csv"
jitter_file = "jitter_results.csv"
results = [f for f in get_files(results_dir) if 'server' not in f]
b_results = [f for f in results if 'jitter' not in f]
j_results = [f for f in results if 'jitter' in f]
df = pd.DataFrame(columns=['platform', 'sent_bps', 'received_bps', 'cpu_client', 'cpu_server'])
for offset, f in enumerate(b_results):
selector = get_platform_selector(f)
iterations = read_json_file(path.join(results_dir, f))
for i, iteration in enumerate(iterations):
end = iteration['end']
sent_bps = end['sum_sent']['bits_per_second']
received_bps = end['sum_received']['bits_per_second']
cpu_local = end['cpu_utilization_percent']['host_total']
cpu_remote = end['cpu_utilization_percent']['remote_total']
index = offset * 10 + i
# Store from the platform perspecvive, hence column "sent_bps" means bits sent by the server, not client
df.loc[index] = [selector, received_bps, sent_bps, cpu_local, cpu_remote]
df.to_csv(path.join(output_dir, bandwidth_file), index=False)
df = pd.DataFrame(columns=['platform', 'jitter_ms', 'bps', 'lost_percent', 'cpu_client', 'cpu_server'])