I recently migrated my server in Nuernberg to Wien. However, I noticed significant lag while connecting to the server via ssh and https. So I did some tests to compare the two servers in Nuernberg and Wien respectively.
I used speedtest-cli to show the connectivity to the internet of my machines. And ran ping and iperf to benchmark the connection. The hostnames of my machines are:
-
mys-karlsruhe-0: Previous Root Server from Netcup in Nuernberg, Germany (RS 1000 G9.5 3M Rabatt)- Note: Please ignore the wrong hostname. I had mistaken it for quite long time.
-
mys-wien-0: New Root Server from Netcup in Wien, Germany (RS 1000 G11 iv 12M) -
mys-pc-arch-0: A desktop PC in Frankfurt am Main, Germany, which connects to Telekom with VDHL Supervectoring 35b -
mys-pc-arch-1: A desktop PC in Trieste, Italy, which connects to Telecom Italia with FTTH- Note: Telekom Italia has no IPv6 connectivity.
-
mys-nyc3-0: A VPS from Digital Ocean in New York, US (DC: NYC3) -
muc-mys-0: A desktop PC in Beijing, China
Here's the result:
From the result of speedtest-cli, the performance of RS in Wien is worse than the RS in Nuernberg, and is far worst than the $4/month VPS from digitalocean. This also does not meet the 2.5Gbps promise in the product description. Ping is also nearly 3 times higher than the one in Neurnberg, and is 8 times to the one from digitalocean.
Among other results, except UDP uplink transmission test, in all tests the new VPS in Wien has much worse performance.
The lag of connection is clearly observable by human. When I ran ssh -vvv there is a significant lag after before debug3: set_sock_tos: set socket 3 IPV6_TCLASS 0x48 and debug1: Connection established. which feels even longer than connecting to New York or Beijing.
Have anyone of you experienced similar problem? How did you address it?
If any netcup stuff want to investigate this, I also did traceroute from different locations. I can share them if they may help.
