piler benchmarks

piler benchmarks

test command #2: smtp-source-new -c -s 20 -m 10000 -l 100000 127.0.0.1:25
                 100 kB email with no attachments (no external helper cmd to run)

environment: VC1S (2 cpu, 2 GB mem) VM at scaleway

a) only piler-smtp is running:

sent 10,000 * 100 kB messages using 20 threads within the same VM

2:28 -> ~4000 messages/min =~ 67 msg/sec

tc #2 => 2:23

VC1M instance:
Jun  3 14:22:51
Jun  3 14:24:48

tc #2 => 1:57


VC1L instance:
Jun  3 14:38:38
Jun  3 14:40:35



b) both piler and piler-smtp are running

sent 10,000 * 100 kB messages using 20 threads within the same VM

4:12 -> 2380 messages/min (only receiving this many emails!)

piler with 2 children took a while to process 10k messages (without attachments): 10:02
so the total throughput is ~1000 msg/sec

tc #2 =>  8:01

VC1M instance:
Jun  3 14:27:45
Jun  3 14:30:22
tc #2: 2:37

piler-smtp: 14:27:45 - 14:29:49
piler: 14:27:46 - 14:30:22

VC1L instance:

A)
piler-smtp: 14:41:25 - 14:43:21 => 1:56
piler: 14:43:32 => 2:07

B) 14:46:04 - 14:48:10  => 2:06
piler-smtp:  14:47:59 => 1:55


-------------------
Piler-ee (notenant)

smtp-source-new -c -s 20 -m 10000 -l 100000 -t archive@worker0 127.0.0.1:25

A) VC1L instance, 7 piler child processes:
#1: 15:14:04 - 15:16:23 => 2:19 = 4260 msg/s
#2: 15:18:40 - 15:20:48 => 2:08 = 4680 msg/s
#3: 15:22:22 - 15:24:23 => 2:01 = 4920 msg/s

B) VC1M instance, 5 piler child processes:
#1: 16:34:29 - 16:42:34
#2: 16:43:42 - 16:50:58
#3: 16:52:58 -

17:15:46 - 17:21:42
17:26:05 - 17:31:45

C) VC1S instance, 3 piler child processes:
#1: 15:31:41 - 15:35:52 => 4:11 = 2340 msg/s
#2: 15:36:54 - 15:40:59 => 4:05 = 2400 msg/s
#3: 15:43:19 - 15:47:22 => 4:03 = 2460 msg/s (with improved mysql settings)