-
Notifications
You must be signed in to change notification settings - Fork 0
/
process_middleware_timers_experiment_gets.py
249 lines (202 loc) · 8.86 KB
/
process_middleware_timers_experiment_gets.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
"""
ASL project - fall 2017
author: Jovan Nikolic
Processes logs generated by middleware
"""
import numpy as np
import csv
import math
class TimerStruct:
def __init__(self):
self.worker_id = -1
self.command = "none"
self.number_of_keys = -1
self.queue_size = -1
self.request_received_time = -1
self.put_in_queue_time = -1
self.taken_out_of_queue_time = -1
self.sent_to_server_time = -1
self.received_complete_response_time = -1
self.response_sent_to_client_time = -1
self.dump_to_disk_flag = -1
path_base_gets = "data/experiment_gets/middleware"
agg_path_base = "aggregated_data/experiment_gets/"
plot_base_path = "plots/experiment_gets/"
client_threads_basename = "clientThreads_"
worker_threads_basename = "_workerThreads_"
counters_basename = "counter_"
timers_basename = "timers_"
number_of_middlewares = 2
repetitions = 3
step = 1e9
command_type = "_S0-G10"
cpt = 2
wt = 64
num_keys = [1, 3, 6, 9]
suffixes = ["sharded", "nonsharded"]
metrics = ["mean", "std"]
all_data = {}
buckets = []
def create_buckets():
b = 0
bucket_step = 0.1
while True:
b += bucket_step
if round(b, 2) >= 15:
break
buckets.append(round(b, 2))
def get_bucket(original_time):
rounded_time = round(original_time, 1)
if rounded_time - original_time >= 0:
return rounded_time
else:
return round(rounded_time + 0.1, 1)
def read_experiments():
for suffix in suffixes:
all_data[suffix] = {}
for keys in num_keys:
all_data[suffix][keys] = {}
for rep in range(repetitions):
print("Read " + suffix + " keys = " + str(keys) + " rep = " + str(rep))
all_data[suffix][keys][rep] = []
current_rep = rep + 1
for mw in range(1, number_of_middlewares + 1):
for worker in range(wt):
path = path_base_gets + str(mw) + "/" + "clientThreads_" + str(cpt) + \
"_workerThreads_" + str(wt) + command_type + "_rep" + str(current_rep) + \
"_" + suffix + "_keys" + str(keys) + "/logs/timers_" + str(worker) + ".log"
with open(path, 'r') as timer_file:
timer_data = timer_file.readlines()
timer_data = [x.strip() for x in timer_data]
for k, line in enumerate(timer_data):
if k == 0:
continue
parsed_line = line.split(',')
[x.strip() for x in parsed_line]
ts = TimerStruct()
ts.worker_id = int(parsed_line[0])
ts.command = parsed_line[1]
ts.number_of_keys = int(parsed_line[2])
ts.queue_size = int(parsed_line[3])
ts.request_received_time = int(parsed_line[4])
ts.put_in_queue_time = int(parsed_line[5])
ts.taken_out_of_queue_time = int(parsed_line[6])
ts.sent_to_server_time = int(parsed_line[7])
ts.received_complete_response_time = int(parsed_line[8])
ts.response_sent_to_client_time = int(parsed_line[9])
ts.dump_to_disk_flag = int(parsed_line[10])
all_data[suffix][keys][rep].append(ts)
def process_for_histograms():
hist_data_respt = {}
hist_data_sst = {}
hist_data_wpt = {}
for suffix in suffixes:
hist_data_respt[suffix] = {}
hist_data_sst[suffix] = {}
hist_data_wpt[suffix] = {}
for keys in num_keys:
hist_data_respt[suffix][keys] = {}
hist_data_sst[suffix][keys] = {}
hist_data_wpt[suffix][keys] = {}
seen_requests = 0
out_of_bucket_requests = 0
for rep in range(repetitions):
hist_data_respt[suffix][keys][rep] = {}
hist_data_sst[suffix][keys][rep] = {}
hist_data_wpt[suffix][keys][rep] = {}
for bucket in buckets:
hist_data_respt[suffix][keys][rep][bucket] = 0
hist_data_sst[suffix][keys][rep][bucket] = 0
hist_data_wpt[suffix][keys][rep][bucket] = 0
for ts in all_data[suffix][keys][rep]:
if ts.command == 'SET':
continue
if ts.number_of_keys != keys:
print("~~~")
continue
seen_requests += 1
res_time = abs(ts.response_sent_to_client_time - ts.request_received_time) / 1e6
ss_time = abs(ts.received_complete_response_time - ts.sent_to_server_time) / 1e6
wp_time = abs(ts.response_sent_to_client_time - ts.received_complete_response_time) / 1e6
if res_time > 7:
print("Flag = " + str(ts.dump_to_disk_flag) + " for response time = " + str(res_time))
b = get_bucket(res_time)
if b not in hist_data_respt[suffix][keys][rep]:
out_of_bucket_requests += 1
else:
hist_data_respt[suffix][keys][rep][b] += 1
b = get_bucket(ss_time)
if b in hist_data_sst[suffix][keys][rep]:
hist_data_sst[suffix][keys][rep][b] += 1
b = get_bucket(wp_time)
if b in hist_data_wpt[suffix][keys][rep]:
hist_data_wpt[suffix][keys][rep][b] += 1
print("Seen requests: " + str(seen_requests) + ", out of bucket was: " + str(out_of_bucket_requests))
print_full_mw_hists(hist_data_respt, "response_time")
print_full_mw_hists(hist_data_sst, "server_service_time")
print_full_mw_hists(hist_data_wpt, "worker_postprocessing_time")
final_data = {}
for suffix in suffixes:
final_data[suffix] = {}
for keys in num_keys:
final_data[suffix][keys] = {}
for b in buckets:
final_data[suffix][keys][b] = {}
jobs = []
for rep in range(repetitions):
jobs.append(hist_data_respt[suffix][keys][rep][b])
mrt = np.mean(np.asarray(jobs))
stdrt = np.std(np.asarray(jobs))
final_data[suffix][keys][b][metrics[0]] = mrt
final_data[suffix][keys][b][metrics[1]] = stdrt
print_final_hists(final_data)
def print_final_hists(final_data):
header = ["#Buckets",
"Mean jobs - Keys 1", "Std jobs - Keys 1",
"Mean jobs - Keys 3", "Std jobs - Keys 3",
"Mean jobs - Keys 6", "Std jobs - Keys 6",
"Mean jobs - Keys 9", "Std jobs - Keys 9"]
for suffix in suffixes:
path = plot_base_path + "histograms_mw_" + suffix + ".csv"
with open(path, 'w') as csv_file:
writer = csv.DictWriter(csv_file, fieldnames=header)
writer.writeheader()
for row in range(len(buckets)):
one_row = {}
i = 0
one_row[header[i]] = buckets[row]
i += 1
for keys in num_keys:
for metric in metrics:
one_row[header[i]] = final_data[suffix][keys][buckets[row]][metric]
i += 1
writer.writerow(one_row)
csv_file.close()
def print_full_mw_hists(hist_data, tag):
header = ["#Buckets",
"Keys 1 - rep 1", "Keys 1 - rep 2", "Keys 1 - rep 3",
"Keys 3 - rep 1", "Keys 3 - rep 2", "Keys 3 - rep 3",
"Keys 6 - rep 1", "Keys 6 - rep 2", "Keys 6 - rep 3",
"Keys 9 - rep 1", "Keys 9 - rep 2", "Keys 9 - rep 3"]
for suffix in suffixes:
path = agg_path_base + "histograms_mw_" + suffix + "_" + tag + ".csv"
with open(path, 'w') as csv_file:
writer = csv.DictWriter(csv_file, fieldnames=header)
writer.writeheader()
for row in range(len(buckets)):
one_row = {}
i = 0
one_row[header[i]] = buckets[row]
i += 1
for keys in num_keys:
for rep in range(repetitions):
one_row[header[i]] = hist_data[suffix][keys][rep][buckets[row]]
i += 1
writer.writerow(one_row)
csv_file.close()
def main():
create_buckets()
read_experiments()
process_for_histograms()
if __name__ == "__main__":
main()