I'm trying to make a simple api application using beego. During the stress test, there was an unexpected problem. Before ~16400 requests everything executes at fantastic speed. After 16400 queries almost all stops, runs 1-2 requests per second. I have a feeling that beego can not allocate a connection to the database. I tried to change maxIdle, maxConn parameters but no effect.
UPD. the same problem with other databases
MainController:
package controllers
import (
models "github.com/Hepri/taxi/models"
"github.com/astaxie/beego"
"github.com/astaxie/beego/orm"
)
type MainController struct {
beego.Controller
}
func (c *MainController) Get() {
o := orm.NewOrm()
app := models.ApiApp{}
err := o.Read(&app)
if err == orm.ErrMissPK {
// do nothing
}
c.ServeJson()
}
Model:
package models
const (
CompanyAccessTypeAll = 1
CompanyAccessTypeSpecific = 2
)
type ApiApp struct {
Id int `orm:"auto"`
Token string `orm:"size(100)"`
}
func (a *ApiApp) TableName() string {
return "api_apps"
}
main.go:
package main
import (
models "github.com/Hepri/taxi/models"
_ "github.com/Hepri/taxi/routers"
"github.com/astaxie/beego"
"github.com/astaxie/beego/orm"
_ "github.com/lib/pq"
)
func main() {
orm.RegisterDriver("postgres", orm.DR_Postgres)
orm.RegisterDataBase("default", "postgres", "user=test password=123456 dbname=test sslmode=disable")
orm.RegisterModel(new(models.ApiApp))
beego.EnableAdmin = true
orm.RunCommand()
beego.Run()
}
before reach ~16400:
Benchmarking localhost (be patient)
^C
Server Software: beegoServer:1.4.2
Server Hostname: localhost
Server Port: 8080
Document Path: /
Document Length: 4 bytes
Concurrency Level: 10
Time taken for tests: 3.844 seconds
Complete requests: 16396
Failed requests: 0
Write errors: 0
Total transferred: 2492192 bytes
HTML transferred: 65584 bytes
Requests per second: 4264.91 [#/sec] (mean)
Time per request: 2.345 [ms] (mean)
Time per request: 0.234 [ms] (mean, across all concurrent requests)
Transfer rate: 633.07 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 0 2.2 0 275
Processing: 0 2 10.9 1 370
Waiting: 0 1 8.6 1 370
Total: 0 2 11.1 2 370
Percentage of the requests served within a certain time (ms)
50% 2
66% 2
75% 2
80% 2
90% 2
95% 3
98% 3
99% 4
100% 370 (longest request)
after reach ~16400:
Benchmarking localhost (be patient)
^C
Server Software: beegoServer:1.4.2
Server Hostname: localhost
Server Port: 8080
Document Path: /
Document Length: 4 bytes
Concurrency Level: 10
Time taken for tests: 15.534 seconds
Complete requests: 16392
Failed requests: 0
Write errors: 0
Total transferred: 2491584 bytes
HTML transferred: 65568 bytes
Requests per second: 1055.22 [#/sec] (mean)
Time per request: 9.477 [ms] (mean)
Time per request: 0.948 [ms] (mean, across all concurrent requests)
Transfer rate: 156.63 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 0 0.3 0 11
Processing: 0 2 16.7 1 614
Waiting: 0 1 15.7 1 614
Total: 0 2 16.7 1 614
Percentage of the requests served within a certain time (ms)
50% 1
66% 1
75% 2
80% 2
90% 2
95% 2
98% 3
99% 3
100% 614 (longest request)
same picture even after 30 seconds
Benchmarking localhost (be patient)
^C
Server Software: beegoServer:1.4.2
Server Hostname: localhost
Server Port: 8080
Document Path: /
Document Length: 4 bytes
Concurrency Level: 10
Time taken for tests: 25.585 seconds
Complete requests: 16391
Failed requests: 0
Write errors: 0
Total transferred: 2491432 bytes
HTML transferred: 65564 bytes
Requests per second: 640.65 [#/sec] (mean)
Time per request: 15.609 [ms] (mean)
Time per request: 1.561 [ms] (mean, across all concurrent requests)
Transfer rate: 95.10 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 1 10.1 0 617
Processing: 0 2 16.2 1 598
Waiting: 0 1 11.1 1 597
Total: 0 2 19.1 1 618
Percentage of the requests served within a certain time (ms)
50% 1
66% 2
75% 2
80% 2
90% 2
95% 2
98% 3
99% 3
100% 618 (longest request)
Related
I was running an UDP test and I noticed that the jitter value was way too high, is something not initialized properly in iperf3 source code? The connection between client and server is very good.
Maybe the reason for high jitter was that prev_transit is not initialized to zero, but I am not sure.
How jitter should work:
http://toncar.cz/Tutorials/VoIP/VoIP_Basics_Jitter.html
Client:
[ 4] local 10.131.136.133 port 49402 connected to 10.131.138.232 port 5201
[ ID] Interval Transfer Bandwidth Total Datagrams
[ 4] 0.00-1.00 sec 16.0 KBytes 131 Kbits/sec 2
[ 4] 1.00-2.00 sec 8.00 KBytes 65.5 Kbits/sec 1
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bandwidth Jitter Lost/Total
Datagrams [ 4]
0.00-2.00 sec 24.0 KBytes 98.2 Kbits/sec 63.064 ms 0/3 (0%)
[ 4] Sent 3 datagrams
iperf Done.
Server:
Starting Test: protocol: UDP, 1 streams, 8192 byte blocks, omitting 0 seconds, 2 second test
[ ID] Interval Transfer Bandwidth Jitter Lost/Total Datagrams
[ 5] 0.00-1.00 sec 16.0 KBytes 131 Kbits/sec 67.261 ms 0/2 (0%)
[ 5] 1.00-2.00 sec 8.00 KBytes 65.5 Kbits/sec 63.064 ms 0/1 (0%)
[ 5] 2.00-2.04 sec 0.00 Bytes 0.00 bits/sec 63.064 ms 0/0 (-nan%)
- - - - - - - - - - - - - - - - - - - - - - - - -
Test Complete. Summary Results:
[ ID] Interval Transfer Bandwidth Jitter Lost/Total Datagrams
[ 5] 0.00-2.04 sec 0.00 Bytes 0.00 bits/sec 63.064 ms 0/3 (0%)
CPU Utilization: local/receiver 0.0% (0.0%u/0.0%s), remote/sender 1.9% (0.3%u/1.8%s)
iperf 3.1
I'm guessing you are the same person who filed this issue in the iperf3 issue tracker because the wording of this question and the one in the issue tracker are almost identical:
https://github.com/esnet/iperf/issues/672
I answered there that you had too few packets per measurement interval to actually compute the jitter in a meaningful way. I suggested that you send at a higher bitrate to get more data points to measure the jitter. Also you should use a version of iperf3 that is 3.2 or newer because of improvements in the timing of sending packets.
I use Locust, a load testing framework, and the following is the summary of a test result.
Name # reqs # fails Avg Min Max | Median req/s
--------------------------------------------------------------------------------------------------------------------------------------------
GET /sample 10000 0(0.00%) 97 56 349 | 96 761.90
--------------------------------------------------------------------------------------------------------------------------------------------
Total 10000 0(0.00%) 761.90
I guess that req/s means 761.90 requests are processed in 1 second. How about Avg, Min, Max and Median? How can I read these columns?.
This perfomance test tooks about 15 sec. I set min_wait = 0 max_wait = 0.
Looking at the source, it appears to refer to the response time
I've got several CentOS 6 servers, synced to pool.ntp.org time-servers.
But sometimes time on them is out of sync, which make difference for 20-30 seconds, which causes errors in my app.
What can be the cause of this, and where should I look for it?
Config
tinker panic 1000 allan 1500 dispersion 15 step 0.128 stepout 900
statsdir /var/log/ntpstats/
leapfile /etc/ntp.leapseconds
driftfile /var/lib/ntp/ntp.drift
statistics loopstats peerstats clockstats
filegen loopstats file loopstats type day enable
filegen peerstats file peerstats type day enable
filegen clockstats file clockstats type day enable
disable monitor
server 0.pool.ntp.org iburst minpoll 6 maxpoll 10
restrict 0.pool.ntp.org nomodify notrap noquery
server 1.pool.ntp.org iburst minpoll 6 maxpoll 10
restrict 1.pool.ntp.org nomodify notrap noquery
server 2.pool.ntp.org iburst minpoll 6 maxpoll 10
restrict 2.pool.ntp.org nomodify notrap noquery
server 3.pool.ntp.org iburst minpoll 6 maxpoll 10
restrict 3.pool.ntp.org nomodify notrap noquery
restrict default kod notrap nomodify nopeer noquery
restrict 127.0.0.1 nomodify
restrict -6 default kod notrap nomodify nopeer noquery
restrict -6 ::1 nomodify
server 127.127.1.0 # local clock
fudge 127.127.1.0 stratum 10
srv1
remote refid st t when poll reach delay offset jitter
==============================================================================
server01.coloce .STEP. 16 u 11d 1024 0 0.000 0.000 0.000
+mt4.raqxs.net 193.190.230.66 2 u 510 1024 377 6.367 5.984 7.433
+16-164-ftth.ons 193.79.237.14 2 u 217 1024 375 11.339 -0.028 4.564
*services.freshd 213.136.0.252 2 u 419 1024 377 6.735 2.048 4.321
LOCAL(0) .LOCL. 10 l - 64 0 0.000 0.000 0.000
srv2
remote refid st t when poll reach delay offset jitter
==============================================================================
+ntp2.edutel.nl 80.94.65.10 2 u 527 1024 377 11.924 1.469 0.753
-95.211.224.12 193.67.79.202 2 u 364 1024 377 12.989 4.930 0.628
+app.kingsquare. 193.79.237.14 2 u 339 1024 377 5.485 0.493 0.591
*ntp.bserved.nl 193.67.79.202 2 u 206 1024 377 7.007 0.539 0.420
LOCAL(0) .LOCL. 10 l - 64 0 0.000 0.000 0.000
I am trying to get a feel for what I should expect in terms of performance from cloud storage.
I just ran the gsutil perfdiag from a compute engine instance in the same location (US) and the same project as my cloud storage bucket.
For nearline storage, I get a 25 Mibit/s read and 353 Mibit/s write, is that low / high / average, why such discrepancy between read and write ?
==============================================================================
DIAGNOSTIC RESULTS
==============================================================================
------------------------------------------------------------------------------
Latency
------------------------------------------------------------------------------
Operation Size Trials Mean (ms) Std Dev (ms) Median (ms) 90th % (ms)
========= ========= ====== ========= ============ =========== ===========
Delete 0 B 5 112.0 52.9 78.2 173.6
Delete 1 KiB 5 94.1 17.5 90.8 115.0
Delete 100 KiB 5 80.4 2.5 79.9 83.4
Delete 1 MiB 5 86.7 3.7 88.2 90.4
Download 0 B 5 58.1 3.8 57.8 62.2
Download 1 KiB 5 2892.4 1071.5 2589.1 4111.9
Download 100 KiB 5 1955.0 711.3 1764.9 2814.3
Download 1 MiB 5 2679.4 976.2 2216.2 3869.9
Metadata 0 B 5 69.1 57.0 42.8 129.3
Metadata 1 KiB 5 37.4 1.5 37.1 39.0
Metadata 100 KiB 5 64.2 47.7 40.9 113.0
Metadata 1 MiB 5 45.7 9.1 49.4 55.1
Upload 0 B 5 138.3 21.0 122.5 164.8
Upload 1 KiB 5 170.6 61.5 139.4 242.0
Upload 100 KiB 5 387.2 294.5 245.8 706.1
Upload 1 MiB 5 257.4 51.3 228.4 319.7
------------------------------------------------------------------------------
Write Throughput
------------------------------------------------------------------------------
Copied a 1 GiB file 5 times for a total transfer size of 5 GiB.
Write throughput: 353.13 Mibit/s.
------------------------------------------------------------------------------
Read Throughput
------------------------------------------------------------------------------
Copied a 1 GiB file 5 times for a total transfer size of 5 GiB.
Read throughput: 25.16 Mibit/s.
------------------------------------------------------------------------------
System Information
------------------------------------------------------------------------------
IP Address:
##.###.###.##
Temporary Directory:
/tmp
Bucket URI:
gs://pl_twitter/
gsutil Version:
4.12
boto Version:
2.30.0
Measurement time:
2015-05-11 07:03:26 PM
Google Server:
Google Server IP Addresses:
##.###.###.###
Google Server Hostnames:
Google DNS thinks your IP is:
CPU Count:
4
CPU Load Average:
[0.16, 0.05, 0.06]
Total Memory:
14.38 GiB
Free Memory:
11.34 GiB
TCP segments sent during test:
5592296
TCP segments received during test:
2417850
TCP segments retransmit during test:
3794
Disk Counter Deltas:
disk reads writes rbytes wbytes rtime wtime
sda1 31 5775 126976 1091674112 856 1603544
TCP /proc values:
wmem_default = 212992
wmem_max = 212992
rmem_default = 212992
tcp_timestamps = 1
tcp_window_scaling = 1
tcp_sack = 1
rmem_max = 212992
Boto HTTPS Enabled:
True
Requests routed through proxy:
False
Latency of the DNS lookup for Google Storage server (ms):
2.5
Latencies connecting to Google Storage server IPs (ms):
##.###.###.### = 1.1
------------------------------------------------------------------------------
In-Process HTTP Statistics
------------------------------------------------------------------------------
Total HTTP requests made: 94
HTTP 5xx errors: 0
HTTP connections broken: 0
Availability: 100%
For standard storage I get:
==============================================================================
DIAGNOSTIC RESULTS
==============================================================================
------------------------------------------------------------------------------
Latency
------------------------------------------------------------------------------
Operation Size Trials Mean (ms) Std Dev (ms) Median (ms) 90th % (ms)
========= ========= ====== ========= ============ =========== ===========
Delete 0 B 5 121.9 34.8 105.1 158.9
Delete 1 KiB 5 159.3 58.2 126.0 232.3
Delete 100 KiB 5 106.8 17.0 103.3 125.7
Delete 1 MiB 5 167.0 77.3 145.1 251.0
Download 0 B 5 87.2 10.3 81.1 100.0
Download 1 KiB 5 95.5 18.0 92.4 115.6
Download 100 KiB 5 156.7 20.5 155.8 179.6
Download 1 MiB 5 219.6 11.7 213.4 232.6
Metadata 0 B 5 59.7 4.5 57.8 64.4
Metadata 1 KiB 5 61.0 21.8 49.6 85.4
Metadata 100 KiB 5 55.3 10.4 50.7 67.7
Metadata 1 MiB 5 75.6 27.8 67.4 109.0
Upload 0 B 5 162.7 37.0 139.0 207.7
Upload 1 KiB 5 165.2 23.6 152.3 194.1
Upload 100 KiB 5 392.1 235.0 268.7 643.0
Upload 1 MiB 5 387.0 79.5 340.9 486.1
------------------------------------------------------------------------------
Write Throughput
------------------------------------------------------------------------------
Copied a 1 GiB file 5 times for a total transfer size of 5 GiB.
Write throughput: 515.63 Mibit/s.
------------------------------------------------------------------------------
Read Throughput
------------------------------------------------------------------------------
Copied a 1 GiB file 5 times for a total transfer size of 5 GiB.
Read throughput: 123.14 Mibit/s.
------------------------------------------------------------------------------
System Information
------------------------------------------------------------------------------
IP Address:
10.240.133.190
Temporary Directory:
/tmp
Bucket URI:
gs://test_throughput_standard/
gsutil Version:
4.12
boto Version:
2.30.0
Measurement time:
2015-05-21 11:08:50 AM
Google Server:
Google Server IP Addresses:
##.###.##.###
Google Server Hostnames:
Google DNS thinks your IP is:
CPU Count:
8
CPU Load Average:
[0.28, 0.18, 0.08]
Total Memory:
Upload 1 MiB 5 387.0 79.5 340.9 486.1
49.91 GiB
Free Memory:
47.9 GiB
TCP segments sent during test:
5165461
TCP segments received during test:
1881727
TCP segments retransmit during test:
3423
Disk Counter Deltas:
disk reads writes rbytes wbytes rtime wtime
dm-0 0 0 0 0 0 0
loop0 0 0 0 0 0 0
loop1 0 0 0 0 0 0
sda1 0 4229 0 1080618496 0 1605286
TCP /proc values:
wmem_default = 212992
wmem_max = 212992
rmem_default = 212992
tcp_timestamps = 1
tcp_window_scaling = 1
tcp_sack = 1
rmem_max = 212992
Boto HTTPS Enabled:
True
Requests routed through proxy:
False
Latency of the DNS lookup for Google Storage server (ms):
1.2
Latencies connecting to Google Storage server IPs (ms):
##.###.##.### = 1.3
------------------------------------------------------------------------------
In-Process HTTP Statistics
------------------------------------------------------------------------------
Total HTTP requests made: 94
HTTP 5xx errors: 0
HTTP connections broken: 0
Availability: 100%
==============================================================================
DIAGNOSTIC RESULTS
==============================================================================
------------------------------------------------------------------------------
Latency
------------------------------------------------------------------------------
Operation Size Trials Mean (ms) Std Dev (ms) Median (ms) 90th % (ms)
========= ========= ====== ========= ============ =========== ===========
Delete 0 B 5 145.1 59.4 117.8 215.2
Delete 1 KiB 5 178.0 51.4 190.6 224.3
Delete 100 KiB 5 98.3 5.0 96.6 104.3
Delete 1 MiB 5 117.7 19.2 112.0 140.2
Download 0 B 5 109.4 38.9 91.9 156.5
Download 1 KiB 5 149.5 41.0 141.9 192.5
Download 100 KiB 5 106.9 20.3 108.6 127.8
Download 1 MiB 5 121.1 16.0 112.2 140.9
Metadata 0 B 5 70.0 10.8 76.8 79.9
Metadata 1 KiB 5 113.8 36.6 124.0 148.7
Metadata 100 KiB 5 63.1 20.2 55.7 86.5
Metadata 1 MiB 5 59.2 4.9 61.3 62.9
Upload 0 B 5 127.5 22.6 117.4 153.6
Upload 1 KiB 5 215.2 54.8 221.4 270.4
Upload 100 KiB 5 229.8 79.2 171.6 329.8
Upload 1 MiB 5 489.8 412.3 295.3 915.4
------------------------------------------------------------------------------
Write Throughput
------------------------------------------------------------------------------
Copied a 1 GiB file 5 times for a total transfer size of 5 GiB.
Write throughput: 503 Mibit/s.
------------------------------------------------------------------------------
Read Throughput
------------------------------------------------------------------------------
Copied a 1 GiB file 5 times for a total transfer size of 5 GiB.
Read throughput: 1.05 Gibit/s.
------------------------------------------------------------------------------
System Information
------------------------------------------------------------------------------
IP Address:
################
Temporary Directory:
/tmp
Bucket URI:
gs://test_throughput_standard/
gsutil Version:
4.12
boto Version:
2.30.0
Measurement time:
2015-05-21 06:20:49 PM
Google Server:
Google Server IP Addresses:
#############
Google Server Hostnames:
Google DNS thinks your IP is:
CPU Count:
8
CPU Load Average:
[0.08, 0.03, 0.05]
Total Memory:
49.91 GiB
Free Memory:
47.95 GiB
TCP segments sent during test:
4958020
TCP segments received during test:
2326124
TCP segments retransmit during test:
2163
Disk Counter Deltas:
disk reads writes rbytes wbytes rtime wtime
dm-0 0 0 0 0 0 0
loop0 0 0 0 0 0 0
loop1 0 0 0 0 0 0
sda1 0 4202 0 1080475136 0 1610000
TCP /proc values:
wmem_default = 212992
wmem_max = 212992
rmem_default = 212992
tcp_timestamps = 1
tcp_window_scaling = 1
tcp_sack = 1
rmem_max = 212992
Boto HTTPS Enabled:
True
Requests routed through proxy:
False
Latency of the DNS lookup for Google Storage server (ms):
1.6
Latencies connecting to Google Storage server IPs (ms):
############ = 1.3
2nd Run:
==============================================================================
DIAGNOSTIC RESULTS
==============================================================================
------------------------------------------------------------------------------
Latency
------------------------------------------------------------------------------
Operation Size Trials Mean (ms) Std Dev (ms) Median (ms) 90th % (ms)
========= ========= ====== ========= ============ =========== ===========
Delete 0 B 5 91.5 14.0 85.1 106.0
Delete 1 KiB 5 125.4 76.2 91.7 203.3
Delete 100 KiB 5 104.4 15.9 99.0 123.2
Delete 1 MiB 5 128.2 36.0 116.4 170.7
Download 0 B 5 60.2 8.3 63.0 68.7
Download 1 KiB 5 62.6 11.3 61.6 74.8
Download 100 KiB 5 103.2 21.3 110.7 123.8
Download 1 MiB 5 137.1 18.5 130.3 159.8
Metadata 0 B 5 73.4 35.9 62.3 114.2
Metadata 1 KiB 5 55.9 18.1 55.3 75.6
Metadata 100 KiB 5 45.7 11.0 42.5 59.1
Metadata 1 MiB 5 49.9 7.9 49.2 58.8
Upload 0 B 5 128.2 24.6 115.5 158.8
Upload 1 KiB 5 153.5 44.1 132.4 206.4
Upload 100 KiB 5 176.8 26.8 165.1 209.7
Upload 1 MiB 5 277.9 80.2 214.7 378.5
------------------------------------------------------------------------------
Write Throughput
------------------------------------------------------------------------------
Copied a 1 GiB file 5 times for a total transfer size of 5 GiB.
Write throughput: 463.76 Mibit/s.
------------------------------------------------------------------------------
Read Throughput
------------------------------------------------------------------------------
Copied a 1 GiB file 5 times for a total transfer size of 5 GiB.
Read throughput: 184.96 Mibit/s.
------------------------------------------------------------------------------
System Information
------------------------------------------------------------------------------
IP Address:
#################
Temporary Directory:
/tmp
Bucket URI:
gs://test_throughput_standard/
gsutil Version:
4.12
boto Version:
2.30.0
Measurement time:
2015-05-21 06:24:31 PM
Google Server:
Google Server IP Addresses:
####################
Google Server Hostnames:
Google DNS thinks your IP is:
CPU Count:
8
CPU Load Average:
[0.19, 0.17, 0.11]
Total Memory:
49.91 GiB
Free Memory:
47.9 GiB
TCP segments sent during test:
5180256
TCP segments received during test:
2034323
TCP segments retransmit during test:
2883
Disk Counter Deltas:
disk reads writes rbytes wbytes rtime wtime
dm-0 0 0 0 0 0 0
loop0 0 0 0 0 0 0
loop1 0 0 0 0 0 0
sda1 0 4209 0 1080480768 0 1604066
TCP /proc values:
wmem_default = 212992
wmem_max = 212992
rmem_default = 212992
tcp_timestamps = 1
tcp_window_scaling = 1
tcp_sack = 1
rmem_max = 212992
Boto HTTPS Enabled:
True
Requests routed through proxy:
False
Latency of the DNS lookup for Google Storage server (ms):
3.5
Latencies connecting to Google Storage server IPs (ms):
################ = 1.1
------------------------------------------------------------------------------
In-Process HTTP Statistics
------------------------------------------------------------------------------
Total HTTP requests made: 94
HTTP 5xx errors: 0
HTTP connections broken: 0
Availability: 100%
3rd run
==============================================================================
DIAGNOSTIC RESULTS
==============================================================================
------------------------------------------------------------------------------
Latency
------------------------------------------------------------------------------
Operation Size Trials Mean (ms) Std Dev (ms) Median (ms) 90th % (ms)
========= ========= ====== ========= ============ =========== ===========
Delete 0 B 5 157.0 78.3 101.5 254.9
Delete 1 KiB 5 153.5 49.1 178.3 202.5
Delete 100 KiB 5 152.9 47.5 168.0 202.6
Delete 1 MiB 5 110.6 20.4 105.7 134.5
Download 0 B 5 104.4 50.5 66.8 167.6
Download 1 KiB 5 68.1 11.1 68.7 79.2
Download 100 KiB 5 85.5 5.8 86.0 90.8
Download 1 MiB 5 126.6 40.1 100.5 175.0
Metadata 0 B 5 67.9 16.2 61.0 86.6
Metadata 1 KiB 5 49.3 8.6 44.9 59.5
Metadata 100 KiB 5 66.6 35.4 44.2 107.8
Metadata 1 MiB 5 53.9 13.2 52.1 69.4
Upload 0 B 5 136.7 37.1 114.4 183.5
Upload 1 KiB 5 145.5 58.3 116.8 208.2
Upload 100 KiB 5 227.3 37.6 233.3 259.3
Upload 1 MiB 5 274.8 45.2 261.8 328.5
------------------------------------------------------------------------------
Write Throughput
------------------------------------------------------------------------------
Copied a 1 GiB file 5 times for a total transfer size of 5 GiB.
Write throughput: 407.03 Mibit/s.
------------------------------------------------------------------------------
Read Throughput
------------------------------------------------------------------------------
Copied a 1 GiB file 5 times for a total transfer size of 5 GiB.
Read throughput: 629.07 Mibit/s.
------------------------------------------------------------------------------
System Information
------------------------------------------------------------------------------
IP Address:
###############
Temporary Directory:
/tmp
Bucket URI:
gs://test_throughput_standard/
gsutil Version:
4.12
boto Version:
2.30.0
Measurement time:
2015-05-21 06:32:48 PM
Google Server:
Google Server IP Addresses:
################
Google Server Hostnames:
Google DNS thinks your IP is:
CPU Count:
8
CPU Load Average:
[0.11, 0.13, 0.13]
Total Memory:
49.91 GiB
Free Memory:
47.94 GiB
TCP segments sent during test:
5603925
TCP segments received during test:
2438425
TCP segments retransmit during test:
4586
Disk Counter Deltas:
disk reads writes rbytes wbytes rtime wtime
dm-0 0 0 0 0 0 0
loop0 0 0 0 0 0 0
loop1 0 0 0 0 0 0
sda1 0 4185 0 1080353792 0 1603851
TCP /proc values:
wmem_default = 212992
wmem_max = 212992
rmem_default = 212992
tcp_timestamps = 1
tcp_window_scaling = 1
tcp_sack = 1
rmem_max = 212992
Boto HTTPS Enabled:
True
Requests routed through proxy:
False
Latency of the DNS lookup for Google Storage server (ms):
2.2
Latencies connecting to Google Storage server IPs (ms):
############## = 1.6
All things being equal, write performance is generally higher for modern storage systems because of presence of a caching layer between the application disks, that said, what you are seeing is within the expected range for "nearline" storage.
I have observed far superior throughput results when using "standard" storage buckets. Though latency did not improve much. Consider using a "Standard" bucket if your application requires high throughput. If your application is sensitive to latency, then using local storage as a cache (or scratch space) may be the only option.
Here is a snippet from one my experiments on "Standard" buckets:
------------------------------------------------------------------------------
Latency
------------------------------------------------------------------------------
Operation Size Trials Mean (ms) Std Dev (ms) Median (ms) 90th % (ms)
========= ========= ====== ========= ============ =========== ===========
Delete 0 B 10 91.5 12.4 89.0 98.5
Delete 1 KiB 10 96.4 9.1 95.6 105.6
Delete 100 KiB 10 92.9 22.8 85.3 102.4
Delete 1 MiB 10 86.4 9.1 84.1 93.2
Download 0 B 10 54.2 5.1 55.4 58.8
Download 1 KiB 10 83.3 18.7 78.4 94.9
Download 100 KiB 10 75.2 14.5 68.6 92.6
Download 1 MiB 10 95.0 19.7 86.3 126.7
Metadata 0 B 10 33.5 7.9 31.1 44.8
Metadata 1 KiB 10 36.3 7.2 35.8 46.8
Metadata 100 KiB 10 37.7 9.2 36.6 44.1
Metadata 1 MiB 10 116.1 231.3 36.6 136.1
Upload 0 B 10 151.4 67.5 122.9 195.9
Upload 1 KiB 10 134.2 22.4 127.9 149.3
Upload 100 KiB 10 168.8 20.5 168.6 188.6
Upload 1 MiB 10 213.3 37.6 200.2 262.5
------------------------------------------------------------------------------
Write Throughput
------------------------------------------------------------------------------
Copied 5 1 GiB file(s) for a total transfer size of 10 GiB.
Write throughput: 3.46 Gibit/s.
Parallelism strategy: both
------------------------------------------------------------------------------
Write Throughput With File I/O
------------------------------------------------------------------------------
Copied 5 1 GiB file(s) for a total transfer size of 10 GiB.
Write throughput: 3.9 Gibit/s.
Parallelism strategy: both
------------------------------------------------------------------------------
Read Throughput
------------------------------------------------------------------------------
Copied 5 1 GiB file(s) for a total transfer size of 10 GiB.
Read throughput: 7.04 Gibit/s.
Parallelism strategy: both
------------------------------------------------------------------------------
Read Throughput With File I/O
------------------------------------------------------------------------------
Copied 5 1 GiB file(s) for a total transfer size of 10 GiB.
Read throughput: 1.64 Gibit/s.
Parallelism strategy: both
Hope that is helpful.
I am using ubuntu 12, nginx, uwsgi 1.9 with socket, django 1.5.
Config:
[uwsgi]
base_path = /home/someuser/web/
module = server.manage_uwsgi
uid = www-data
gid = www-data
virtualenv = /home/someuser
master = true
vacuum = true
harakiri = 20
harakiri-verbose = true
log-x-forwarded-for = true
profiler = true
no-orphans = true
max-requests = 10000
cpu-affinity = 1
workers = 4
reload-on-as = 512
listen = 3000
Client tests from Windows7:
C:\Users\user>C:\AppServ\Apache2.2\bin\ab.exe -c 255 -n 5000 http://www.someweb.com/about/
This is ApacheBench, Version 2.0.40-dev <$Revision: 1.146 $> apache-2.0
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Copyright 2006 The Apache Software Foundation, http://www.apache.org/
Benchmarking www.someweb.com (be patient)
Completed 500 requests
Completed 1000 requests
Completed 1500 requests
Completed 2000 requests
Completed 2500 requests
Completed 3000 requests
Completed 3500 requests
Completed 4000 requests
Completed 4500 requests
Finished 5000 requests
Server Software: nginx
Server Hostname: www.someweb.com
Server Port: 80
Document Path: /about/
Document Length: 1881 bytes
Concurrency Level: 255
Time taken for tests: 66.669814 seconds
Complete requests: 5000
Failed requests: 1
(Connect: 1, Length: 0, Exceptions: 0)
Write errors: 0
Total transferred: 10285000 bytes
HTML transferred: 9405000 bytes
Requests per second: 75.00 [#/sec] (mean)
Time per request: 3400.161 [ms] (mean)
Time per request: 13.334 [ms] (mean, across all concurrent requests)
Transfer rate: 150.64 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 8 207.8 1 9007
Processing: 10 3380 11480.5 440 54421
Waiting: 6 1060 3396.5 271 48424
Total: 11 3389 11498.5 441 54423
Percentage of the requests served within a certain time (ms)
50% 441
66% 466
75% 499
80% 519
90% 3415
95% 36440
98% 54407
99% 54413
100% 54423 (longest request)
I have set following options too:
echo 3000 > /proc/sys/net/core/netdev_max_backlog
echo 3000 > /proc/sys/net/core/somaxconn
So,
1) I make first 3000 requests super fast. I see progress in ab and in uwsgi requests logs -
[pid: 5056|app: 0|req: 518/4997] 80.114.157.139 () {30 vars in 378 bytes} [Thu Mar 21 12:37:31 2013] GET /about/ => generated 1881 bytes in 4 msecs (HTTP/1.0 200) 3 headers in 105 bytes (1 switches on core 0)
[pid: 5052|app: 0|req: 512/4998] 80.114.157.139 () {30 vars in 378 bytes} [Thu Mar 21 12:37:31 2013] GET /about/ => generated 1881 bytes in 4 msecs (HTTP/1.0 200) 3 headers in 105 bytes (1 switches on core 0)
[pid: 5054|app: 0|req: 353/4999] 80.114.157.139 () {30 vars in 378 bytes} [Thu Mar 21 12:37:31 2013] GET /about/ => generated 1881 bytes in 4 msecs (HTTP/1.0 200) 3 headers in 105 bytes (1 switches on core 0)
I dont have any broken pipes or worker respawns.
2) Next requests are running very slow or with some timeout. Looks like that some buffer becomes full and I am waiting before it becomes empty.
3) Some buffer becomes empty.
4) ~500 requests are processed super fast.
5) Some timeout.
6) see Nr. 4
7) see Nr. 5
8) see Nr. 4
9) see Nr. 5
....
....
Need your help
check with netstat and dmesg. You have probably exhausted ephemeral ports or filled the conntrack table.