[tiwil]

Just ran another test to someone hosting in Sweden. Got download around 20MB/s and upload for more than 50MB/s which is fast.
I believe server should have higher upload since we're uploading file to user's device instead of downloading them. That's why, we, as a user need more download speed.

Result: http://www.speedtest.net/result/10933710976.png

---

When tested from nearby location, this is the speed:

Code: (Select All)

Location                       Upload           Download         Ping
---------------------------------------------------------------------------
Nearby                         901.49 Mbit/s    394.54 Mbit/s    5.174 ms

10MB-20MB/s seems fine for far distance, as it is being tested from USA which is far from Finland. As the ping increases, speed decreases:

Code: (Select All)

USA, New York (Optimum)        160.92 Mbit/s    99.40 Mbit/s     98.398 ms
USA, Chicago (Windstream)      160.60 Mbit/s    216.81 Mbit/s   109.641 ms
USA, Dallas (Frontier)         130.53 Mbit/s    154.86 Mbit/s   144.184 ms
USA, Miami (Frontier)          134.63 Mbit/s    111.45 Mbit/s   141.256 ms
USA, Los Angeles (Windstream)  122.98 Mbit/s    169.06 Mbit/s   161.338 ms

Nearest Location, considering this is on a 1 Gbps Shared Port, this also seems fine.

Code: (Select All)

France, Lyon (SFR)             322.70 Mbit/s    457.59 Mbit/s    47.241 ms
Germany, Berlin (DNS:NET)      376.35 Mbit/s    317.23 Mbit/s    30.758 ms

Well that's fast for nearest location and nearby location. I cannot find France or Germany in my nearest speedtest server list tho. Can you post the server ID so I can try it out?
Probably if I cannot get it up to 300Mbit/s on download probably there is something to do with the networking on my VPS. I don't really know for sure what made that happen.

[Lampard]

You can run the following benchmark and check the speedtest from all around the world: https://bench.monster/
Command:

Code: (Select All)

curl -LsO bench.monster/speedtest.sh; bash speedtest.sh

[tiwil]

You can run the following benchmark and check the speedtest from all around the world: https://bench.monster/
Command:

Code: (Select All)

curl -LsO bench.monster/speedtest.sh; bash speedtest.sh

Thanks @Lampard. I got similar result with your result. Pretty sure it's Speedtest's server that has network speed "problem".
I also realize that it has various result too, like RAM speed, CPU benchmark, and others. Will add it to the opening post after this.

What a perfect VPS!

[tbelldesignco]

Thanks for your suggestion, bro! I appreciate it. I only use mail server for sending email out as I need to setup a no-reply email for a specific domain.

Other than that I use Yandex services as it's free and fast to configure too. Therefore there is no incoming mail.

Plus, this VPS never go down for maintenance from the first time I powered up! Which is fantastic! I usually only reboot the server when Ubuntu asks to.

Right on! Yeah I looked at Yandex when I was looking at SMTP send-only relays, but I decided I needed more functionality. Also that is great you have amazing uptime! I have experienced similar experiences where my VPS will only go down when I update Ubuntu or there are server migrations, which has only had me down for maybe 45minutes at the max.

[tiwil]

Also that is great you have amazing uptime! I have experienced similar experiences where my VPS will only go down when I update Ubuntu or there are server migrations, which has only had me down for maybe 45minutes at the max.

Even better! I announced a planned downtime for 15 minutes to restart the server so Ubuntu's update can take place. But it's like not even restarting and it's already up. I mean, when I restart we got disconnected from the VPS right. Then after that I immediately reconnect and it's already up!

[fChk]

No surprise here really, just like it was predicted in this post.

I think the 3 new VPSs are interesting in their own ways.

> VPS 13 is the most appealing with its NVME-based SSD storage if one has moderate requirements for storage, memory and CPU; I'm assuming that it has 1 vCPU.
(...)

The only correction there is that this VPS has 2 vCPUs instead of the assumed 1, which makes it perfect CPU-wise.

Disk I/O wise, I would think that it's capped at 700/750 MB/s -given the NVMe interface- which is, nevertheless, optimal for most use-cases.

Thus basically I would just cross the CPU in that quote of mine to get something like: "VPS-13 is an ideal VPS for someone who has moderate requirements in RAM and disk storage".

[fChk]

I'm interested in the disk I/O performance of the Post4VPS VPSs with NVMe SSD-backed storage (ie VPS-7, VPS-13, VPS-14.)

Thus can you please run the following commands on your VPS and share the output with us:

Code: (Select All)

# Your block device details
lsblk --output "NAME,KNAME,MODEL,HCTL,SIZE,VENDOR,SUBSYSTEMS"

# Investigating the used kernel drivers (virtio vs nvme)
lsmod | grep virt
lsmod | grep nvm
lshw -class storage
lshw -class disk

# Disk seq.read Speed (Cached)
hdparm -Tt /dev/vda (or /dev/sda)

# Disk seq.read Speed (Direct)
hdparm -Tt --direct /dev/vda (or /dev/sda)

# Disk Latency:
ioping -c 15 /home/mysudoer
ioping -c 15 /dev/vda (or /dev/sda)

# Disk Seek Rate:
ioping -R /dev/vda (or /dev/sda)

# Disk Sequential Speed:
ioping -RL /dev/vda (or /dev/sda)

Finally I'm also interested in the results of the fio Disk Speed Tests (Mixed R/W 50/50), which is included in the "Yet-Another-Bench-Script" available on github:
https://github.com/masonr/yet-another-bench-script

To avoid running all the yab's tests, you can restrict it to the disk speed with the ig flags:

Code: (Select All)

./yabs.sh -ig

Thanks!