Virtono – Cloud Mini (DE) benchmark / review
Hi LET
I had been looking for a small yearly KVM in EU for a while, when I noticed that Virtono were adding a EUR 9.95/y "Cloud Mini" to their offering I decided to give it a try. Most cheap offers I found either had too little disk space or not enough RAM, this little beast (15GB SSD / 1 Core / 512MB RAM / 1 IPv4 / 1 IPv6 / 1TB Bandwidth @ 1Gbps) seemed quite capable.
I did some traceroutes and it seemed to me that Frankfurt was the best location for quick routing from various places in Europe. I was initially planning to downgrade the Miami KVM I have with them, but when I asked them if that was possible a few weeks ago there was no very small KVMs available, but they offered me to try a small OpenVZ server in another location which I wasn't interested in. When I saw that they started offering small KVMs I decided to pull the trigger and try a "Cloud Mini" KVM in Germany.
The Cloud Mini in DE was delivered really quickly. They are using WHMCS and Virtualizor, as I mentioned in my review of their KVM VPS in Miami. Everything works as expected.
Here is a bench.sh:
----------------------------------------------------------------------
CPU model : Intel(R) Xeon(R) CPU X5650 @ 2.67GHz
Number of cores : 1
CPU frequency : 2666.760 MHz
Total size of Disk : 16.0 GB (3.9 GB Used)
Total amount of Mem : 487 MB (86 MB Used)
Total amount of Swap : 1279 MB (15 MB Used)
System uptime : 0 days, 2 hour 36 min
Load average : 2.86, 2.34, 1.44
OS : CentOS 7.6.1810
Arch : x86_64 (64 Bit)
Kernel : 3.10.0-957.10.1.el7.x86_64
----------------------------------------------------------------------
I/O speed(1st run) : 436 MB/s
I/O speed(2nd run) : 373 MB/s
I/O speed(3rd run) : 392 MB/s
Average I/O speed : 400.3 MB/s
----------------------------------------------------------------------
Node Name IPv4 address Download Speed
CacheFly 205.234.175.175 34.4MB/s
Linode, Tokyo, JP 106.187.96.148 5.70MB/s
Linode, Singapore, SG 139.162.23.4 8.51MB/s
Linode, London, UK 176.58.107.39 42.5MB/s
Linode, Frankfurt, DE 139.162.130.8 55.7MB/s
Linode, Fremont, CA 50.116.14.9 3.67MB/s
Softlayer, Dallas, TX 173.192.68.18 12.9MB/s
Softlayer, Seattle, WA 67.228.112.250 12.3MB/s
Softlayer, Frankfurt, DE 159.122.69.4 9.11MB/s
Softlayer, Singapore, SG 119.81.28.170 11.5MB/s
Softlayer, HongKong, CN 119.81.130.170 9.66MB/s
----------------------------------------------------------------------
Node Name IPv6 address Download Speed
Linode, Atlanta, GA 2600:3c02::4b 11.9MB/s
Linode, Dallas, TX 2600:3c00::4b 6.87MB/s
Linode, Newark, NJ 2600:3c03::4b 11.6MB/s
Linode, Singapore, SG 2400:8901::4b 5.41MB/s
Linode, Tokyo, JP 2400:8900::4b 5.28MB/s
Softlayer, San Jose, CA 2607:f0d0:2601:2a::4 10.3MB/s
Softlayer, Paris, FR 2a03:8180:1301:8::4 35.4MB/s
Softlayer, Singapore, SG 2401:c900:1101:8::2 9.00MB/s
Softlayer, Tokyo, JP 2401:c900:1001:16::4 6.98MB/s
----------------------------------------------------------------------
And here is the initial serverscope bench: https://serverscope.io/trials/6qZq
You'll notice on it that the CPU is capped to around 1/4th of it's full power, as UnixBench result is 236.0. When asking the support about that, they told me that all they "Tiny" VPS plans are capped (KVM and OpenVZ) and that users needing more power should go to higher end plans: CLOUD S or SMALL VPS and higher aren't capped.
IMO the ability to use 100% of one core for a few minutes/hours per day would have been better, but their choice is understandable given the very cheap price, and not necessarily a problem, as for many use cases this could be more than enough power. But this has to be taken into consideration before choosing to order one of their 'Tiny' VPS or another option. If your project is to run a busy php site this is probably not the right VPS, but for a small dynamic site or to serve your cached pages with nginx, it should perform decently.
The network seems good to me, the node is in Telehouse Frankfurt and they are using m247 (https://bgp.he.net/AS9009) like in their other locations.
Comments
CPU speed is capped at 1/2 of full CPU power, not 1/4: https://www.lowendtalk.com/discussion/comment/2959997/#Comment_2959997
In case this can be useful to fellow LET members here are more benchmarks (without CPU capping):
nench.sh:
Serverscope:
https://serverscope.io/trials/B9ar
Latency:
Can someone with their service try to do a high PPS transfer and see if they still auto-null-route / suspend you? My experience with their services directly, when previously testing them, was that any high rate PPS transfer would cause the server to be suspended. I am curious to know if this is resolved?
To note: I now have services through their re-seller Greenwebpages and don't see this happen, but per their explanation in their thread, they supposedly have their own hypervisors and manage them a bit differently.
Also, when I tested it was in the DE location.
Appreciate any feedback on this.
Cheers!
What kind of test would work out for you? (What activity did get you null-routed?)
Maybe @catalingmn has a direct answer to your question, though.
Thanks for the bench. How about a new UnixBench with 1/2 Core now?
Cheers
I can't remember exactly, but I would expect my first thought was to check iperf, so very likely iperf would be enough to have triggered what I saw. It has been 6-8 months since I had a server though, so things could have changed for the better.
Cheers!
As I understand it, the first one is with what you can expect from "1/2 core" (1/4 was my guess, based on unixbench results given the CPU used).
Just tried: I had no problem running many iperf tests.
Are you sure?
Reading @catalingmn post, it sounded like an issue which has been resolved...
Yep. I just contacted support, they confirmed that. It was 1/2 from the beginning, 1/4 was just my (wrong) guess. This CPU cap is only on the smaller plans (Cloud Mini / Tiny VPS), all other plans aren't capped at all.
Any review or comment on this provider? Virtono seems formally was shockvps.