All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Dual locations Contabo benchmark & review
First some general remarks not specifically about this particular review:
As soon as one does connectivity benchmarking one meets an ugly beast, namely the fact that the internet is far less perfect than one off testing would suggest and a target host that usually is fast and fine suddenly needs an eternity (as in > half an hour) to respond to a ping or a https request. The result - until now and not any more - was that providers sometimes could look worse than they actually are; even worse they often were innocent victims of some target server going awol.
Well noted I always chose the target servers carefully and in fact they usually worked fine, but not always. Occasionally some LET user would suggest to use other test target because he considered mine (e.g. Softlayer) poor, and trust me, I did try and test many target alternatives. The sad truth seems to be that every test target server occasionally PMSes.
So I thought about a better solution and finally it struck me and I stopped my attempt to find more reliable target servers but rather simply ignored those few bad results (and it's always just a few, less then about 5%). This review is the first one where I applied the new result compiling and calculating method which will be used for all my reviews from now on.
But at the same time one should note that that problem only arose from what actually is a significant advantage of my way of testing and my software; one gets to see things the other benchmarks almost always miss plus the candidate (VPS under test) is tested for multiple days (minimum) and during all times, day and night with the timing being unpredictable.
Now the results of 2 Contabo locations, one in DE the other one in the USA. First the german one, based on over 200 result sets/runs:
Machine: amd64, Arch.: amd64, Model: AMD EPYC 7282 16-Core Processor
OS, version: FreeBSD 12.2, Mem.: 7.982 GB
CPU - Cores: 4, Family/Model/Stepping: 23/49/0
Cache: 64K/64K L1d/L1i, 512K L2, 16M L3
Std. Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat
pse36 cflsh mmx fxsr sse sse2 htt sse3 pclmulqdq ssse3 fma cx16
sse4_1 sse4_2 popcnt *aes* xsave osxsave avx f16c rdrnd hypervisor
Ext. Flags: syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm lahf_lm cmp_legacy
cr8_legacy lzcnt sse4a misalignsse 3dnowprefetch osvw perfctr_core
ProcMem SC [MB/s]: avg 286.5 - min 207.8 (72.5 %), max 326.6 (114.0 %)
ProcMem MC [MB/s]: avg 879.1 - min 536.5 (61.0 %), max 1023.9 (116.5 %)
--- Disk - Buffered ---
Write seq. [MB/s]: avg 320.60 - min 74.16 (23.1%), max 483.13 (150.7%)
Write rnd. [MB/s]: avg 986.30 - min 651.45 (66.0%), max 2234.96 (226.6%)
Read seq. [MB/s]: avg 1802.08 - min 707.83 (39.3%), max 3377.26 (187.4%)
Read rnd. [MB/s]: avg 5355.50 - min 2528.76 (47.2%), max 7874.74 (147.0%)
--- Disk - Sync/Direct ---
Write seq. [MB/s]: avg 120.56 - min 106.74 (88.5%), max 156.15 (129.5%)
Write rnd. [MB/s]: avg 110.59 - min 108.07 (97.7%), max 129.26 (116.9%)
Read seq. [MB/s]: avg 1357.27 - min 550.03 (40.5%), max 3049.95 (224.7%)
Read rnd. [MB/s]: avg 5589.39 - min 2221.76 (39.7%), max 7791.22 (139.4%)
--- Network ---
US LAX lax.download.datapacket.com
DL [Mb/s]: avg 32.02 - min 18.18 (56.8%), max 44.80 (139.9%)
Ping [ms]: avg 150.2 - min 143.4 (95.5%), max 167.1 (111.3%)
Web ping [ms]: avg 151.7 - min 146.8 (96.8%), max 175.7 (115.8%)
NO OSL speedtest.osl01.softlayer.com
DL [Mb/s]: avg 30.10 - min 7.35 (24.4%), max 141.08 (468.8%)
Ping [ms]: avg 35.2 - min 32.8 (93.3%), max 44.9 (127.7%)
Web ping [ms]: avg 35.9 - min 32.8 (91.4%), max 74.3 (207.1%)
US SJC speedtest.sjc01.softlayer.com
DL [Mb/s]: avg 16.21 - min 0.18 (1.1%), max 737.23 (4547.4%)
Ping [ms]: avg 150.3 - min 149.3 (99.3%), max 163.6 (108.8%)
Web ping [ms]: avg 162.0 - min 149.4 (92.2%), max 671.7 (414.7%)
JP TOK speedtest.tokyo2.linode.com
DL [Mb/s]: avg 10.39 - min 3.86 (37.1%), max 27.44 (264.0%)
Ping [ms]: avg 258.1 - min 256.3 (99.3%), max 268.4 (104.0%)
Web ping [ms]: avg 259.7 - min 256.3 (98.7%), max 295.8 (113.9%)
IT MIL speedtest.mil01.softlayer.com
DL [Mb/s]: avg 54.89 - min 14.33 (26.1%), max 178.68 (325.5%)
Ping [ms]: avg 14.4 - min 13.5 (93.8%), max 27.3 (189.6%)
Web ping [ms]: avg 23.6 - min 13.5 (57.2%), max 745.4 (3158.0%)
FR PAR speedtest.par01.softlayer.com
DL [Mb/s]: avg 77.57 - min 16.74 (21.6%), max 180.00 (232.0%)
Ping [ms]: avg 14.5 - min 13.2 (91.1%), max 25.1 (173.3%)
Web ping [ms]: avg 24.6 - min 13.2 (53.7%), max 561.8 (2287.4%)
SG SGP mirror.sg.leaseweb.net
DL [Mb/s]: avg 8.85 - min 0.00 (0.0%), max 26.06 (294.4%) - (http error: -1)
Ping [ms]: avg 346.8 - min 264.6 (76.3%), max 374.3 (107.9%)
Web ping [ms]: avg 351.7 - min 264.6 (75.2%), max 374.6 (106.5%)
BR SAO speedtest.sao01.softlayer.com
DL [Mb/s]: avg 10.34 - min 0.00 (0.0%), max 31.40 (303.5%) - (http error: -10)
Ping [ms]: avg 192.8 - min 191.5 (99.3%), max 209.5 (108.6%)
Web ping [ms]: avg 195.3 - min 191.6 (98.1%), max 375.8 (192.4%)
IN CHN speedtest.che01.softlayer.com
DL [Mb/s]: avg 7.97 - min 2.91 (36.5%), max 23.24 (291.5%)
Ping [ms]: avg 273.2 - min 110.1 (40.3%), max 287.9 (105.4%)
Web ping [ms]: avg 281.0 - min 260.1 (92.6%), max 471.2 (167.7%)
GR UNK speedtest.ftp.otenet.gr
DL [Mb/s]: avg 128.45 - min 127.11 (99.0%), max 129.79 (101.0%)
Ping [ms]: avg 40.0 - min 39.9 (99.9%), max 40.0 (100.1%)
Web ping [ms]: avg 40.5 - min 40.4 (99.6%), max 40.7 (100.4%)
US WDC mirror.wdc1.us.leaseweb.net
DL [Mb/s]: avg 19.22 - min 5.73 (29.8%), max 62.07 (322.9%)
Ping [ms]: avg 99.9 - min 91.7 (91.8%), max 119.1 (119.2%)
Web ping [ms]: avg 100.2 - min 91.8 (91.6%), max 119.1 (118.8%)
DE FRA speedtest.fra02.softlayer.com
DL [Mb/s]: avg 114.39 - min 34.71 (30.3%), max 188.38 (164.7%)
Ping [ms]: avg 5.3 - min 4.4 (83.6%), max 20.9 (396.9%)
Web ping [ms]: avg 7.9 - min 4.5 (56.6%), max 360.2 (4532.8%)
RU MOS speedtest.hostkey.ru
DL [Mb/s]: avg 35.95 - min 2.98 (8.3%), max 139.10 (387.0%)
Ping [ms]: avg 41.5 - min 35.9 (86.5%), max 60.5 (145.8%)
Web ping [ms]: avg 43.7 - min 38.7 (88.6%), max 82.0 (187.7%)
US DAL speedtest.dal05.softlayer.com
DL [Mb/s]: avg 18.32 - min 3.46 (18.9%), max 49.89 (272.4%)
Ping [ms]: avg 130.9 - min 125.1 (95.6%), max 144.2 (110.2%)
Web ping [ms]: avg 131.6 - min 125.2 (95.2%), max 159.5 (121.2%)
UK LON speedtest.lon02.softlayer.com
DL [Mb/s]: avg 99.94 - min 5.88 (5.9%), max 180.83 (180.9%)
Ping [ms]: avg 13.7 - min 11.0 (80.3%), max 20.9 (152.6%)
Web ping [ms]: avg 18.7 - min 11.0 (58.9%), max 596.8 (3194.0%)
US NYC nyc.download.datapacket.com
DL [Mb/s]: avg 53.74 - min 26.71 (49.7%), max 171.97 (320.0%)
Ping [ms]: avg 81.6 - min 12.4 (15.2%), max 96.0 (117.6%)
Web ping [ms]: avg 82.4 - min 12.4 (15.0%), max 118.9 (144.3%)
RO BUC 185.183.99.8
DL [Mb/s]: avg 51.18 - min 12.45 (24.3%), max 141.53 (276.6%)
Ping [ms]: avg 37.7 - min 32.2 (85.3%), max 45.6 (120.9%)
Web ping [ms]: avg 39.8 - min 32.7 (82.1%), max 68.6 (172.2%)
CN_HK mirror.hk.leaseweb.net
DL [Mb/s]: avg 11.00 - min 0.19 (1.7%), max 26.74 (243.1%)
Ping [ms]: avg 282.4 - min 250.6 (88.7%), max 768.2 (272.0%)
Web ping [ms]: avg 282.6 - min 253.9 (89.9%), max 768.2 (271.9%)
4 (!) vCores of a current Epyc! 8 GB memory and 200 GB (!) SSD!
The processor performance is really nice; the only processor achieving a better result so far is Ryzen 39xx/59xx which is about 20% faster. I'm also quite happy with the multi-core result. Really, really nice hardware which isn't brutally oversold.
Regarding the disk I'm a bit split. On the one hand both peak and average performance is really decent, even good, but on the other hand the spread indicates a controller with poor caching and/or to many users per disk. I'd love to see the load being spread over more disk devices (yes, I'm talking about hardware devices). Normally I wouldn't mention that because the performance is certainly with a region normal for low to mid end VPS, but with such a nice system I' a bit disappointed to see what highly likely a few large devices instead of many smaller ones. (Remember: With Epyc PCIe lanes are not scarce but Sata/Sas 6Gb is a limit).
As for the network I'm happy with what I see; I'd call that solid mid range. Keep in mid that (at least officially) that VPS is limited to 200 Mb/s - which is plenty enough for most users, especially in the low end segment. The spread seems to be a bit on the higher end though and I'd like to see a bit more consistency.
(... part 2 ...)
Comments
that's pretty good! even disk isn't bad at all to me
Now to the US location, based on over 150 result sets/runs:
The processor and memory are the same as above but there is considerably less spread and the performance is better too.
On the other hand the disk results are much worse and so is the spread; in fact, quite a few test runs barely worked.
As for the network, first 2 remarks: Their pipe provider seems to dislike southern Europe. Almost all failures where ti southern european targets, one of which I even had to remove because almost all connection attempts had absolutely horrible latency.
Secondly, The size of this result set is much smaller than the one for the DE location because Contabo (USA) had a network problem and I decided to throw away all results collected so far because they wouldn't show results under normal conditions.
The results are OK to decent and in fact among the better sets I've seen so far in/from the USA.
Now to some points that are not benchmarked or expressable in numbers but nevertheless noteworthy; in fact, some of the following might well be decisive for your decision to buy or not to buy.
Most of it turns around the Contabo panel. It's not bad and some might even like it (I for example did like it) but it's seriously different from the panels one is used to and it's somewhat limited. The one weakness that probably is the major turn off for many potential clients, especially less experienced ones, is that there is a way to VNC into your box, even with nice bells and jingles like setting your own password - but use must use a desktop VNC viewer; there is no viewer in the panel.
Somewhat similarly, Contabo's support seems to be good, professional, and friendly - but it's via email, there is no ticket system in the panel.
Summary: Very nice nodes with fast processors and memory and mostly decent connectivity, but the disks, especially in the USA location are not up to par. Plus their panel is not what most users are used to and, together with the lack of a ticket system in the panel, might turn off quite a few customers.
On the bright side you get a really attractive VPS with 4 fast vCores, plenty memory, and a large SSD disk for an almost insanely low price (€ 4.99/mo).
TL;DR If you are willing and capable to walk a bit outside of the main stream track you'll get an all in all really nice and attractive VPS for a very low price.
Have you noticed that the I/O steal is high? It is always hoovering around 30-50% all the time:
avg-cpu: %user %nice %system %iowait %steal %idle
............................. 2.68 0.15 1.09 0.80 30.16 65.12
@jsg Thank you for an extensive benchmark. I'm happy you have found the performance of our VPS to be pretty good overall. I have passed your feedback regarding control panel, hardware and networking issues to the relevant teams. It seems that some of these things can be resolved fairly easily so I hope that some of these quirks will disappear and the VPS will become even better.
I've been using Contabo for 3 months now. With CPU steal usually around 25-35% I am not convinced that the nodes aren't brutally oversold.
I'm using their SSD M VPS, and it's been okay so far.
Their customer support is indeed really good, however they sometimes fail to understand basic english sentences which has caused me a few inconveniences. Or they're just so overworked that they don't read email carefully... Besides that, I don't mind using email instead of live chat, it makes it more organized and better to follow up on. I can also refer to previous emails if I wanted to.
Also note, that in my experience (I have used their Xeon and AMD VPSs as well), the Xeon VPS's cpu performance seems to me 60% worse than it's AMD counterpart.
My VPS L SSD, i've been with Contabo since May 2020. CPU Steal is usually 0.0-0.1%. First had the VPS S SSD, then later upgraded to a VPS L SSD. Later upgraded to AMD Epyc and saw a huge performance increase.
I raised a ticket with them and I was fairly pleasant. They answered the same day with some tweaks and once we saw no difference they moved the VM to another host which fixed the high I/O steal so for now thumbs up especially for this price.
Does Contabo has any policy cancellation similar to netcup or I can cancel anytime?
How do they compare to netcup reliability and performance?
Nope, contabo doesn't have questionable billing practices. Just cancel service from client area. Done that multiple times. No problems.
Thanks for helping, I would have needed to contact them to ask because I got the VPS for free (and returned them back/"cancelled" them once I was done testing).
I use Contabo as VPN server. CPU usage are always under 2%. Guess im a good client.
Contabo is good, but the way they run their business you have to be patient.
The cancellation is painless. done from the CP.
Their prices for lowend VPS have very little competition. KVM with lots of Memory and SSD Disk space, DDoS Protection, 32 TB of transfer, 200 Mbit/s Port....
You get free IPv6's
The Bad:
*With them, you have to order the server when you don't need it immediately.
*You make a order and it take forever to get set-up.
*The price of extras IPv4 too expensive, €3.00
*Control Panel is old as fuck!
If you don't pay for one year in advance you have to pay for setup.
If you want your VPS in US -Missouris that's a extras €1.00 for lowend VPS.
From my personal experience:
Once the server was setup I have not problems.
I will order their server with patient!
My 2 cents!
As per their marketing material, they usually setup server within 20 minutes after payment. They charge extra 2 EUR for USA location if server is bigger, seems it is percent of server price. And of course that "setup fee" (can be waived out with longer contract).
Confirmed, that matches my experience with Contabo.
Great reviews. ThNks!