ASTF rand IP distribution between client & servers

Petr Redkin

unread,

Apr 21, 2021, 1:58:49 PM4/21/21

to TRex Traffic Generator

Hello,

I want to achieve predictible random in distribution between client & server IP ranges but option distribution with "rand" in ASTFIPGenDist doesn't help - I see same traffic as with "seq" with 1 to 1 distribution between client & server IP.

ip_gen_c = ASTFIPGenDist(ip_range=["172.16.0.2", "172.16.3.234"], distribution="rand")

ip_gen_s = ASTFIPGenDist(ip_range=["8.1.0.109", "8.1.3.93"], distribution="rand")

ip_gen = ASTFIPGen(glob=ASTFIPGenGlobal(ip_offset="1.0.0.0"), dist_client=ip_gen_c, dist_server=ip_gen_s)

I achieved what I need when changed only last octet of server to value 81,

ip_gen_s = ASTFIPGenDist(ip_range=["8.1.0.109", "8.1.3.81"], distribution="rand")

but

1 – with value 85 I achieve the same 1 to 1 distribution

2 – with value 94 I achieve pseudo-random distribution

Is this some type of a bug? May I predict value of last octet to always achieve random distribution or in another way to get it?

ASTF @ v2.89

Thanks.

hanoh haim

unread,

Apr 22, 2021, 9:06:26 AM4/22/21

to Petr Redkin, TRex Traffic Generator

Hi Petr,

It depends on how many cores and how many ports you have. The range of the server is split equally to each core.

I've took your range in case of an issue (2133) and simulate 1-21 number of cores and I don't see the issue.

Could you check with 2 ports and one core (-c 1)

j=0
--------------
0
280
1611
978
1136
467
100
1448
1448
1993
j=1
--------------
408
553
885
36
56
564
715
8
408
71
j=2
--------------
296
488
418
661
601
374
65
464
295
498
j=3
--------------
485
406
139
25
392
174
337
403
528
194
j=4
--------------
105
418
307
320
277
30
269
376
116
185
j=5
--------------
272
169
84
97
127
59
172
318
322
21
j=6
--------------
275
153
156
96
299
150
80
27
288
22
j=7
--------------
133
102
73
243
140
123
250
13
202
204
j=8
--------------
196
29
3
163
205
149
174
171
236
210
j=9
--------------
49
65
74
109
125
180
87
179
57
88
j=10
--------------
103
90
55
34
29
110
154
6
103
96
j=11
--------------
169
132
98
157
110
149
28
37
126
23
j=12
--------------
14
45
0
67
4
116
153
39
29
52
j=13
--------------
134
99
22
103
58
58
75
22
89
128
j=14
--------------
83
135
78
21
139
58
20
80
35
69
j=15
--------------
61
127
16
26
42
83
16
86
82
106
j=16
--------------
30
59
48
25
3
112
53
17
118
51
j=17
--------------
15
104
10
19
8
43
29
15
92
53
j=18
--------------
39
50
90
104
72
24
76
101
28
96
j=19
--------------
49
53
63
86
80
48
100
67
46
87

--
You received this message because you are subscribed to the Google Groups "TRex Traffic Generator" group.
To unsubscribe from this group and stop receiving emails from it, send an email to trex-tgn+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/trex-tgn/6728ed38-ee96-4e49-bd78-ffdd5f143e82n%40googlegroups.com.

--

Hanoh

Sent from my iPhone

hanoh haim

unread,

Apr 22, 2021, 9:18:45 AM4/22/21

to Petr Redkin, TRex Traffic Generator

Hi Petr,

I read again the question, you probably **sorting* the clients and looks for the servers right?

We are using the same random generator for both client and server ranges, so this is expected if the range is the same.

If your range (in each core) is the same the random stream will be the same because the seed is the same (and predictable).

e.g. for core=1

client range - 0..10

server range - 0..10

you will get the same stream of random hence 1:1 relation betwean them,

In the other example you gave there is a different ranges.

The best is to give a different range and give a prime number range per core.

total-range = cores * prime_number

so if you have 20 cores total (2 ports and -c =20)

client range = 20*13

server range = 20 * 97

Thanks

Hanoh

Petr Redkin

unread,

Apr 23, 2021, 4:06:37 AM4/23/21

to TRex Traffic Generator

Thank you Hanoh!

For now I statically assigned client & server ranges which produce required distribution (not 1 client IP always connecting to 1 server IP).

ip_gen_c = ASTFIPGenDist(ip_range=[172.18.0.0, 172.18.3.255], distribution="rand")

ip_gen_s = ASTFIPGenDist(ip_range=[8.1.0.109, 8.1.3.255], distribution="rand")