Performance Issues
2018-03-19 16:48:33.684078+00 by
Dan Lyke
7 comments
Hmmm...
time openssl s_client -CApath /etc/ssl/certs/ -connect www.flutterby.com:443 -
showcerts <<< ""
takes .1 seconds on my server, .577 seconds from home (weird? Maybe I don't have the CA
chain on the home server?), but it takes ~1.7 seconds to serve the static index.html on
both the server and from the home machine.
WTF is Apache doing with those extra 1.1+ seconds?
[ related topics:
Free Software Open Source
]
comments in ascending chronological order (reverse):
#Comment Re: Performance Issues made: 2018-03-19 18:42:45.361648+00 by:
brainopener
[edit history]
If I execute this:
time curl https://www.flutterby.com/index.html > /dev/null
I get this:
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 41289 100 41289 0 0 41289 0 0:00:01 --:--:-- 0:00:01 61350
real 0m0.697s
user 0m0.023s
sys 0m0.009s
So much less than 1700 milliseconds.
This is over wifi to a cable modem in sleepy beach town Florida.
#Comment Re: Performance Issues made: 2018-03-19 22:38:53.767869+00 by:
Dan Lyke
Huh. On the one hand, that's good (though 2/3rds of a second still seems excessive), on the
other hand I wonder what's up on my connection...
#Comment Re: Performance Issues made: 2018-03-19 23:08:29.730958+00 by:
TheSHAD0W
What's the ping to the server?
#Comment Re: Performance Issues made: 2018-03-20 12:23:54.511718+00 by:
brainopener
[edit history]
Some more data...
ARIN says RIPE says that Flutterby IP is French.
Ping from home in Florida:
ping -c 3 www.flutterby.com
PING scaleways.flutterby.com (212.47.229.131): 56 data bytes
64 bytes from 212.47.229.131: icmp_seq=0 ttl=41 time=114.987 ms
64 bytes from 212.47.229.131: icmp_seq=1 ttl=41 time=113.822 ms
64 bytes from 212.47.229.131: icmp_seq=2 ttl=41 time=115.784 ms
--- scaleways.flutterby.com ping statistics ---
3 packets transmitted, 3 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 113.822/114.864/115.784/0.806 ms
Ping from Oregon data center:
ping -c 3 www.flutterby.com
PING scaleways.flutterby.com (212.47.229.131) 56(84) bytes of data.
64 bytes from scaleways.flutterby.net (212.47.229.131): icmp_seq=1 ttl=33 time=171 ms
64 bytes from scaleways.flutterby.net (212.47.229.131): icmp_seq=2 ttl=33 time=171 ms
64 bytes from scaleways.flutterby.net (212.47.229.131): icmp_seq=3 ttl=33 time=171 ms
--- scaleways.flutterby.com ping statistics ---
3 packets transmitted, 3 received, 0% packet loss, time 3939ms
rtt min/avg/max/mdev = 171.855/171.857/171.860/0.478 ms
That same curl command above from Oregon datacenter:
time curl https://www.flutterby.com/index.html > /dev/null
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 44431 100 44431 0 0 36281 0 0:00:01 0:00:01 --:--:-- 36299
real 0m1.231s
user 0m0.068s
sys 0m0.048s
Traceroute from home:
traceroute www.flutterby.com
traceroute to scaleways.flutterby.com (212.47.229.131), 64 hops max, 52 byte packets
1 192.168.118.1 (192.168.118.1) 2.420 ms 1.305 ms 1.217 ms
2 10.55.64.1 (10.55.64.1) 6.422 ms 5.195 ms 8.130 ms
3 ten0-0-0-4.orld39-ser1.bhn.net (72.31.216.150) 6.806 ms 7.965 ms 7.689 ms
4 ten0-6-0-3.orld11-car1.bhn.net (72.31.217.130) 11.756 ms 6.933 ms 10.826 ms
5 72-31-220-174.net.bhntampa.com (72.31.220.174) 10.438 ms
72-31-220-180.net.bhntampa.com (72.31.220.180) 10.053 ms
72-31-220-174.net.bhntampa.com (72.31.220.174) 10.886 ms
6 72-31-220-228.res.bhn.net (72.31.220.228) 10.884 ms
72-31-217-88.net.bhntampa.com (72.31.217.88) 10.352 ms
72-31-197-138.net.bhntampa.com (72.31.197.138) 10.830 ms
7 10.bu-ether15.orldfljo00w-bcr00.tbone.rr.com (66.109.6.98) 12.762 ms 15.245 ms 12.407 ms
8 ae-1.bar2.orlando1.level3.net (4.68.111.17) 20.007 ms 19.148 ms 17.773 ms
9 * * *
10 212.3.235.202 (212.3.235.202) 119.821 ms 117.135 ms 114.699 ms
11 49-infra-dc2-a9k2.dc3.poneytelecom.eu (195.154.1.119) 109.671 ms 112.889 ms 109.691 ms
12 * * *
13 * * *
14 * * *
15 * * *
16 scaleways.flutterby.net (212.47.229.131) 113.262 ms 113.926 ms 116.205 ms
Traceroute from Oregon:
traceroute www.flutterby.com
traceroute to www.flutterby.com (212.47.229.131), 30 hops max, 60 byte packets
1 ec2-50-112-0-150.us-west-2.compute.amazonaws.com (50.112.0.150) 40.895 ms ec2-50-112-0-144.us-
west-2.compute.amazonaws.com (50.112.0.144) 13.431 ms ec2-50-112-0-208.us-west-
2.compute.amazonaws.com (50.112.0.208) 22.396 ms
2 100.66.8.128 (100.66.8.128) 19.995 ms 100.66.8.118 (100.66.8.118) 15.930 ms 100.66.8.4 (100.66.8.4)
13.985 ms
3 100.66.11.196 (100.66.11.196) 14.185 ms 100.66.10.32 (100.66.10.32) 12.194 ms 100.66.11.130
(100.66.11.130) 20.399 ms
4 100.66.7.7 (100.66.7.7) 14.771 ms 100.66.7.197 (100.66.7.197) 16.479 ms 100.66.7.141 (100.66.7.141)
14.189 ms
5 100.66.4.25 (100.66.4.25) 15.831 ms 100.66.4.233 (100.66.4.233) 21.955 ms 100.66.4.135 (100.66.4.135)
14.501 ms
6 100.65.9.33 (100.65.9.33) 0.509 ms 100.65.11.97 (100.65.11.97) 6.368 ms 100.65.8.65 (100.65.8.65)
0.377 ms
7 52.93.240.100 (52.93.240.100) 1.765 ms 52.93.15.230 (52.93.15.230) 0.623 ms 52.93.15.224
(52.93.15.224) 0.644 ms
8 52.93.12.210 (52.93.12.210) 27.437 ms 52.93.12.138 (52.93.12.138) 13.393 ms 12.240 ms
9 52.93.12.187 (52.93.12.187) 0.800 ms 52.93.12.85 (52.93.12.85) 0.855 ms 52.93.12.45 (52.93.12.45)
0.752 ms
10 100.91.25.244 (100.91.25.244) 5.163 ms 100.91.25.128 (100.91.25.128) 5.019 ms 100.91.25.178
(100.91.25.178) 5.135 ms
11 54.239.43.137 (54.239.43.137) 4.486 ms 54.239.45.128 (54.239.45.128) 4.645 ms 54.239.46.103
(54.239.46.103) 4.638 ms
12 52.95.40.132 (52.95.40.132) 4.687 ms 52.95.40.84 (52.95.40.84) 7.308 ms 15.557 ms
13 52.95.40.41 (52.95.40.41) 4.511 ms 52.95.40.57 (52.95.40.57) 4.495 ms 52.95.40.91 (52.95.40.91)
4.560 ms
14 port-b1-link.telia.net (213.248.96.4) 7.283 ms 7.205 ms 52.95.40.107 (52.95.40.107) 4.447 ms
15 palo-b22-link.telia.net (62.115.115.24) 23.716 ms 23.689 ms 23.782 ms
16 nyk-bb3-link.telia.net (62.115.114.4) 83.455 ms ash-bb3-link.telia.net (62.115.122.43) 93.562 ms 93.548
ms
17 prs-bb4-link.telia.net (80.91.251.101) 154.533 ms prs-bb4-link.telia.net (62.115.122.158) 156.136 ms prs-
bb3-link.telia.net (80.91.251.242) 176.329 ms
18 prs-b8-link.telia.net (62.115.138.139) 157.034 ms prs-b8-link.telia.net (62.115.138.133) 167.175 ms prs-
b8-link.telia.net (62.115.138.139) 155.457 ms
19 online-ic-315748-prs-b8.c.telia.net (62.115.63.94) 182.612 ms 181.070 ms 169.483 ms
20 195.154.1.189 (195.154.1.189) 178.913 ms 167.707 ms 178.641 ms
21 * * *
22 * * *
23 * * *
24 * * *
25 scaleways.flutterby.net (212.47.229.131) 178.485 ms 182.639 ms 171.911 ms
#Comment Re: Performance Issues made: 2018-03-20 13:05:02.658496+00 by:
DaveP
[edit history]
81 ~/ % time curl https://www.flutterby.com/index.html > /dev/null
8:02am
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 44431 100 44431 0 0 35297 0 0:00:01 0:00:01 --:--:-- 35290
0.013u 0.008s 0:01.26 0.7% 0+0k 0+5io 0pf+0w
Seems pretty reasonable. My work network connects to the internet in a PacBell building in San Jose.
83 ~/ % ping www.flutterby.com 8:03am
PING scaleways.flutterby.com (212.47.229.131): 56 data bytes
64 bytes from 212.47.229.131: icmp_seq=0 ttl=48 time=207.114 ms
64 bytes from 212.47.229.131: icmp_seq=1 ttl=48 time=207.928 ms
64 bytes from 212.47.229.131: icmp_seq=2 ttl=48 time=207.832 ms
64 bytes from 212.47.229.131: icmp_seq=3 ttl=48 time=208.015 ms
That's a little pokey, but I am in Minneapolis and going through eight hops before I get to SJC where it hits
the
greater internet (and takes about 60ms to ping the host there).
#Comment Re: Performance Issues made: 2018-03-20 13:38:37.242701+00 by:
brainopener
I see that it is ~7,200 miles between Ormond by the Sea, FL and Paris as the pelican flies.
My arithmetic says that it takes light ~40 milliseconds to travel that far.
So that's my base case. It would appear that most of my ping time (~80 ms) is lost to physics.
#Comment Re: Performance Issues made: 2018-03-20 17:48:33.670763+00 by:
Dan Lyke
Yeah, the server's in france. And I guess by the time the full SSL handshake happens,
that's what it is.
I was kind of wondering if I had one of those partial keychain issues going on. I also keep
thinking that re-implementing the site in Lighttpd or some server where I'm a little more
sure there's less ridiculous overhead going on would be a good thing...