Server Upgrade


EvilDragon

Administrator
Staff member
Joined
Mar 4, 2003
Messages
29,986
Age
46
Location
Ingolstadt
It's time to upgrade the server once again.
Besides getting a faster CPU, I also want to move the databases to NVMe-disks to speed up everything and also increase the overall storage.

My service provider (Hetzner) now also offers servers in Finland.

As I know that some of you (US, Australia, etc.) have the issue of slow response times, I wanted to check upfront whether the server in Finland would improve things or not :)

All you need to do is run a ping to the old server and the new one in Finland.

So for the potential new one:

Code:
ping hel.icmp.hetzner.com

and:

Code:
ping openpandora.org

for the old one.

My results here are:

Code:
PING hel.icmp.hetzner.com (95.217.255.75) 56(84) bytes of data. 
64 bytes from hel.icmp.hetzner.com (95.217.255.75): icmp_seq=1 ttl=54 time=45.0 ms 
64 bytes from hel.icmp.hetzner.com (95.217.255.75): icmp_seq=2 ttl=54 time=44.4 ms 
64 bytes from hel.icmp.hetzner.com (95.217.255.75): icmp_seq=3 ttl=54 time=44.5 ms 
64 bytes from hel.icmp.hetzner.com (95.217.255.75): icmp_seq=4 ttl=54 time=43.9 ms

PING openpandora.org (88.99.27.96) 56(84) bytes of data. 
64 bytes from kronos.dragonbox.de (88.99.27.96): icmp_seq=1 ttl=55 time=27.1 ms 
64 bytes from kronos.dragonbox.de (88.99.27.96): icmp_seq=2 ttl=55 time=26.4 ms 
64 bytes from kronos.dragonbox.de (88.99.27.96): icmp_seq=3 ttl=55 time=26.2 ms 
64 bytes from kronos.dragonbox.de (88.99.27.96): icmp_seq=4 ttl=55 time=26.5 ms

So for Germany, response time would be a bit slower.
If things improve overall worldwide though, I'd get the server in Finland (it's a bit cheaper as well).
 
Here in the Netherlands it is about twice as bad (29ms for the Finnish one and 15ms for the current one).
 
From Spain:

Code:
Pinging hel.icmp.hetzner.com [95.217.255.75] with 32 bytes of data:
Reply from 95.217.255.75: bytes=32 time=68ms TTL=56
Reply from 95.217.255.75: bytes=32 time=68ms TTL=56
Reply from 95.217.255.75: bytes=32 time=68ms TTL=56
Reply from 95.217.255.75: bytes=32 time=68ms TTL=56

Ping statistics for 95.217.255.75:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 68ms, Maximum = 68ms, Average = 68ms

C:\>ping openpandora.org

Pinging openpandora.org [88.99.27.96] with 32 bytes of data:
Reply from 88.99.27.96: bytes=32 time=53ms TTL=57
Reply from 88.99.27.96: bytes=32 time=53ms TTL=57
Reply from 88.99.27.96: bytes=32 time=53ms TTL=57
Reply from 88.99.27.96: bytes=32 time=53ms TTL=57

Ping statistics for 88.99.27.96:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 53ms, Maximum = 53ms, Average = 53ms
 
From East Coast US

Y:\>ping hel.icmp.hetzner.com

Pinging hel.icmp.hetzner.com [95.217.255.75] with 32 bytes of data:
Reply from 95.217.255.75: bytes=32 time=107ms TTL=50
Reply from 95.217.255.75: bytes=32 time=107ms TTL=50
Reply from 95.217.255.75: bytes=32 time=108ms TTL=50
Reply from 95.217.255.75: bytes=32 time=107ms TTL=50

Ping statistics for 95.217.255.75:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 107ms, Maximum = 108ms, Average = 107ms

Y:\>ping openpandora.org

Pinging openpandora.org [88.99.27.96] with 32 bytes of data:
Reply from 88.99.27.96: bytes=32 time=92ms TTL=52
Reply from 88.99.27.96: bytes=32 time=92ms TTL=52
Reply from 88.99.27.96: bytes=32 time=92ms TTL=52
Reply from 88.99.27.96: bytes=32 time=92ms TTL=52

Ping statistics for 88.99.27.96:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 92ms, Maximum = 92ms, Average = 92ms
 
For me (NL):
hel.icmp.hetzner.com [95.217.255.75] 50ms
openpandora.org [88.99.27.96] 36ms

Not sure if I would notice 14ms though, if the transfer of data itself would be faster it might be faster in general.
 
From Italy:

Ping hel.icmp.hetzner.com [95.217.255.75] con 32 byte di dati:
Risposta da 95.217.255.75: byte=32 durata=68ms TTL=49
Risposta da 95.217.255.75: byte=32 durata=73ms TTL=49
Risposta da 95.217.255.75: byte=32 durata=78ms TTL=49
Risposta da 95.217.255.75: byte=32 durata=58ms TTL=49

Tempo approssimativo percorsi andata/ritorno in millisecondi:
Minimo = 58ms, Massimo = 78ms, Medio = 69ms


Ping openpandora.org [88.99.27.96] con 32 byte di dati:
Risposta da 88.99.27.96: byte=32 durata=49ms TTL=50
Risposta da 88.99.27.96: byte=32 durata=38ms TTL=50
Risposta da 88.99.27.96: byte=32 durata=48ms TTL=50
Risposta da 88.99.27.96: byte=32 durata=59ms TTL=50


Tempo approssimativo percorsi andata/ritorno in millisecondi:
Minimo = 38ms, Massimo = 59ms, Medio = 48ms

so Finland servers are a bit slower.
 
UK:

Code:
Pinging hel.icmp.hetzner.com [95.217.255.75] with 32 bytes of data:
Reply from 95.217.255.75: bytes=32 time=52ms TTL=51
Reply from 95.217.255.75: bytes=32 time=53ms TTL=51
Reply from 95.217.255.75: bytes=32 time=53ms TTL=51
Reply from 95.217.255.75: bytes=32 time=53ms TTL=51

Ping statistics for 95.217.255.75:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 52ms, Maximum = 53ms, Average = 52ms

Pinging openpandora.org [88.99.27.96] with 32 bytes of data:
Reply from 88.99.27.96: bytes=32 time=37ms TTL=50
Reply from 88.99.27.96: bytes=32 time=38ms TTL=50
Reply from 88.99.27.96: bytes=32 time=38ms TTL=50
Reply from 88.99.27.96: bytes=32 time=38ms TTL=50

Ping statistics for 88.99.27.96:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 37ms, Maximum = 38ms, Average = 37ms
 
US west coast. they're about the same for me. i don't have too much of an issue with response time, in the first place, though...

Code:
ping hel.icmp.hetzner.com
PING hel.icmp.hetzner.com(hel.icmp.hetzner.com (2a01:4f9:0:a010::add:1a)) 56 data bytes
64 bytes from hel.icmp.hetzner.com (2a01:4f9:0:a010::add:1a): icmp_seq=1 ttl=40 time=182 ms
[...]
--- hel.icmp.hetzner.com ping statistics ---
11 packets transmitted, 11 received, 0% packet loss, time 10009ms
rtt min/avg/max/mdev = 179.996/213.521/266.330/30.021 ms


ping openpandora.org
PING openpandora.org(2a01:4f8:10a:39f::b (2a01:4f8:10a:39f::b)) 56 data bytes
64 bytes from 2a01:4f8:10a:39f::b (2a01:4f8:10a:39f::b): icmp_seq=1 ttl=44 time=178 ms
[...]
--- openpandora.org ping statistics ---
11 packets transmitted, 11 received, 0% packet loss, time 10015ms
rtt min/avg/max/mdev = 174.563/213.103/265.908/33.639 ms
 
Germany, so basically useless information, yay:
13ms openpandora.org
28ms hel.icmp.hetzner.com

Since that's obviously too slow, I'd recommend putting the server closer to me. Some services manage 8ms, so please.


On a more serious note, I've been getting random database errors from time to time when browsing for a few weeks now. Nothing that wasn't resolved by just refreshing. Is the reason known? Perhaps it'll automagically be fixed by an upgraded server, who knows.
 
~42 to Finland
~61 to Germany
1459672476466.png

Edit: From the north of Norway
 
Last edited:
Seattle, WA, US:

Ping statistics for 95.217.255.75:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 160ms, Maximum = 160ms, Average = 160ms

Ping statistics for 88.99.27.96:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 145ms, Maximum = 145ms, Average = 145ms
 
US West: hel.icmp.hetzner.com ~193ms, openpandora.org ~177ms

I remember the boards being very slow a few years ago, but they've been good since 2016 or so. Looking forward to the repo site running a bit faster after the server upgrade!
 
From Washington, USA:

Code:
silent@hpmini:~$ ping hel.icmp.hetzner.com
PING hel.icmp.hetzner.com (95.217.255.75) 56(84) bytes of data.
64 bytes from hel.icmp.hetzner.com (95.217.255.75): icmp_seq=1 ttl=49 time=197 ms
64 bytes from hel.icmp.hetzner.com (95.217.255.75): icmp_seq=2 ttl=49 time=190 ms
64 bytes from hel.icmp.hetzner.com (95.217.255.75): icmp_seq=3 ttl=49 time=188 ms
64 bytes from hel.icmp.hetzner.com (95.217.255.75): icmp_seq=4 ttl=49 time=206 ms

silent@hpmini:~$ ping openpandora.org
PING openpandora.org (88.99.27.96) 56(84) bytes of data.
64 bytes from kronos.dragonbox.de (88.99.27.96): icmp_seq=1 ttl=41 time=176 ms
64 bytes from kronos.dragonbox.de (88.99.27.96): icmp_seq=2 ttl=41 time=179 ms
64 bytes from kronos.dragonbox.de (88.99.27.96): icmp_seq=3 ttl=41 time=181 ms
64 bytes from kronos.dragonbox.de (88.99.27.96): icmp_seq=4 ttl=41 time=179 ms

The new one will be a bit slower.
 
From Houston Texas:

Hetzner:

Ping statistics for 2a01:4f9:0:a010::add:1a:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 156ms, Maximum = 163ms, Average = 158ms

Current:

Ping statistics for 2a01:4f8:10a:39f::b:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 142ms, Maximum = 147ms, Average = 144ms
 
East Coast, Australia:

Code:
Finland:
Approximate round trip times in milli-seconds:
Minimum = 397ms, Maximum = 398ms, Average = 397ms

Germany:
Approximate round trip times in milli-seconds:
Minimum = 366ms, Maximum = 369ms, Average = 366ms
 
North East USA
Code:
trashy@Trashy:~$ ping hel.icmp.hetzner.com
PING hel.icmp.hetzner.com (95.217.255.75) 56(84) bytes of data.
64 bytes from hel.icmp.hetzner.com (95.217.255.75): icmp_seq=1 ttl=50 time=128 ms
64 bytes from hel.icmp.hetzner.com (95.217.255.75): icmp_seq=2 ttl=50 time=123 ms
64 bytes from hel.icmp.hetzner.com (95.217.255.75): icmp_seq=3 ttl=50 time=126 ms
64 bytes from hel.icmp.hetzner.com (95.217.255.75): icmp_seq=4 ttl=50 time=128 ms
64 bytes from hel.icmp.hetzner.com (95.217.255.75): icmp_seq=5 ttl=50 time=123 ms
64 bytes from hel.icmp.hetzner.com (95.217.255.75): icmp_seq=6 ttl=50 time=124 ms
64 bytes from hel.icmp.hetzner.com (95.217.255.75): icmp_seq=7 ttl=50 time=128 ms
64 bytes from hel.icmp.hetzner.com (95.217.255.75): icmp_seq=8 ttl=50 time=123 ms
64 bytes from hel.icmp.hetzner.com (95.217.255.75): icmp_seq=9 ttl=50 time=124 ms
64 bytes from hel.icmp.hetzner.com (95.217.255.75): icmp_seq=10 ttl=50 time=129 ms
64 bytes from hel.icmp.hetzner.com (95.217.255.75): icmp_seq=11 ttl=50 time=126 ms
64 bytes from hel.icmp.hetzner.com (95.217.255.75): icmp_seq=12 ttl=50 time=124 ms
64 bytes from hel.icmp.hetzner.com (95.217.255.75): icmp_seq=13 ttl=50 time=123 ms
^C
--- hel.icmp.hetzner.com ping statistics ---
13 packets transmitted, 13 received, 0% packet loss, time 12015ms
rtt min/avg/max/mdev = 123.036/125.650/129.404/2.327 ms
trashy@Trashy:~$ ping openpandora.org
PING openpandora.org (88.99.27.96) 56(84) bytes of data.
64 bytes from kronos.dragonbox.de (88.99.27.96): icmp_seq=1 ttl=50 time=193 ms
64 bytes from kronos.dragonbox.de (88.99.27.96): icmp_seq=2 ttl=50 time=189 ms
64 bytes from kronos.dragonbox.de (88.99.27.96): icmp_seq=3 ttl=50 time=227 ms
64 bytes from kronos.dragonbox.de (88.99.27.96): icmp_seq=4 ttl=50 time=211 ms
64 bytes from kronos.dragonbox.de (88.99.27.96): icmp_seq=5 ttl=50 time=231 ms
64 bytes from kronos.dragonbox.de (88.99.27.96): icmp_seq=6 ttl=50 time=213 ms
64 bytes from kronos.dragonbox.de (88.99.27.96): icmp_seq=7 ttl=50 time=186 ms
64 bytes from kronos.dragonbox.de (88.99.27.96): icmp_seq=8 ttl=50 time=206 ms
64 bytes from kronos.dragonbox.de (88.99.27.96): icmp_seq=9 ttl=50 time=186 ms
64 bytes from kronos.dragonbox.de (88.99.27.96): icmp_seq=10 ttl=50 time=181 ms
64 bytes from kronos.dragonbox.de (88.99.27.96): icmp_seq=11 ttl=50 time=210 ms
64 bytes from kronos.dragonbox.de (88.99.27.96): icmp_seq=12 ttl=50 time=208 ms
64 bytes from kronos.dragonbox.de (88.99.27.96): icmp_seq=13 ttl=50 time=199 ms
^C
--- openpandora.org ping statistics ---
13 packets transmitted, 13 received, 0% packet loss, time 12016ms
rtt min/avg/max/mdev = 181.078/203.516/231.567/15.237 ms
 
On a more serious note, I've been getting random database errors from time to time when browsing for a few weeks now. Nothing that wasn't resolved by just refreshing. Is the reason known? Perhaps it'll automagically be fixed by an upgraded server, who knows.

That's because I had to limit the connections otherwise it would be overloaded.
The error always appears when no connections are free anymore.

With the database being on an NVMe instead of a harddisk in the new server, queries should be faster and therefore, connections shouldn't get to the limit that fast.
 
  • Like
Reactions: rSl
Confirming numbers from Melbourne, Australia. Negligible increase.
Code:
Pinging hel.icmp.hetzner.com [95.217.255.75] with 32 bytes of data:
    Minimum = 348ms, Maximum = 349ms, Average = 348ms

Pinging openpandora.org [88.99.27.96] with 32 bytes of data:
    Minimum = 330ms, Maximum = 332ms, Average = 330ms
 
Back
Top