Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!


Big mivo benchmarks/review - 4 locations
New on LowEndTalk? Please Register and read our Community Rules.

All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.

Big mivo benchmarks/review - 4 locations

jsgjsg Member, Resident Benchmarker
edited November 2019 in Reviews

Let me start with a big "Thank you!" to mivo who provided me with access to a VPS in 4 locations: Moldavia, Romania, Russia, and USA. I changed the formatting slighty so that you can see the probably most interesting value for you (Avg.) more clearly.

That leads me directly to the first point I liked: all VPSs had the same CPU, an Ivy Bridge E3-12xx, and 2 GB memory. This makes the VPSs nicely comparable and makes life easier for the mivo people which again leads to happy customers.

Here's the full System Info, once only for all because as I said they are basically the same:

Machine: amd64, Arch.: amd64, Model: Intel Xeon E3-12xx v2 (Ivy Bridge, IBRS)
OS, version: FreeBSD 12.0, Mem.: 1.985 GB
CPU - Cores: 2, Family/Model/Stepping: 6/58/9
Cache: 32K/32K L1d/L1i, 2M 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 ss sse3 pclmulqdq ssse3 cx16 pcid
          sse4_1 sse4_2 x2apic popcnt tsc_deadline aes xsave osxsave avx f16c
          rdrnd hypervisor
Ext. Flags: fsgsbase tsc_adjust smep erms syscall nx pdpe1gb rdtscp lm lahf_lm

Nice CPU and the VPS has all the interesting flags available. I also call it nice because it has lots of L2 and L3 cache.

Processor & memory MD/RO/RU:

                Avg       -        Low High Dlow Dhigh Span
single core:       241.03   -       230.29  252.62  4.45    4.81    9.26
multi core:   513.04    -       279.30  561.38  45.56   9.42    54.98

and USA

single core:   313.98   -       302.95  326.65  3.51    4.04    7.55
multi core:   729.58    -       694.79  761.84  4.77    4.42    9.19

While there are minor differences between the MD/RO/RU systems those are easily explained by different users, different load, etc. The fact that they are so close to each other indicates that mivo doesn't tolerate abusers and keeps an eye on their nodes. Note though that those nodes are under normal load as the multi-core span suggests. But the values are still good.

The US node(s) seem to have a newer and/or considerably higher clocked processor version. a single core value is clearly in the E-26xx range so those VPSs might be a particularly attractive choice for US-Americans or people needing a VPS there.

Disk:

RO and RU
Seq.Wr: 402.32  -       38.92   456.40  90.33   13.44   103.77
Rnd.Wr: 55.36   -       38.09   81.57   31.20   47.35   78.54
Seq.Rd: 689.29  -       84.68   843.80  87.71   22.41   110.13
Rnd.Rd: 709.80  -       43.01   869.03  93.94   22.43   116.37

MD
Seq.Wr: 191.56  -       164.20  232.82  14.28   21.54   35.82
Rnd.Wr: 61.42   -       54.74   94.20   10.88   53.37   64.25
Seq.Rd: 352.65  -       288.38  476.08  18.23   35.00   53.23
Rnd.Rd: 501.97  -       386.03  626.27  23.10   24.76   47.86

USA
Seq.Wr: 550.78  -       5.44    587.66  99.01   6.70    105.71
Rnd.Wr: 53.12   -       6.60    58.25   87.57   9.66    97.24
Seq.Rd: 919.85  -       393.16  1024.00 57.26   11.32   68.58
Rnd.Rd: 985.52  -       787.77  1114.00 20.07   13.04   33.10

In fact, the romanian VPS had a slightly faster disk than the russian one they are very similar so I put just the russian results here. I want to note however that the romanian disk has a considerably lower span, meaning it behaved considerably more consistent while the russian disk seems to be under quite a bit of stress. But still, the results are good and show solid good SSD values.
As a reminder: my random writing benchmark really does random writes; both the data and the access pattern are purely random and hence trash caches (which I want them to do), so don't be put off by the low values.

The disk on the moldavian node is considerably slower and probably not an SSD (or a rather old and slow one). But still, seq. writing with almost 200 MB/s is not seriously bad (but may look so in todays SSD universe).

The USA node again stands out positively. I guess it's simply newer and uses a more current hardware generation. Nice results!

Network:

Moldavia
GR_UNK: 26.61   -       1.70    47.30   93.61   77.76   171.37
US_SJC: 17.15   -       2.20    29.50   87.17   72.03   159.20
OK_LON: 42.40   -       4.10    77.20   90.33   82.06   172.39
DE_FRA: 52.57   -       3.50    82.80   93.34   57.50   150.84
BR_SAO: 20.64   -       18.70   22.50   9.40    9.01    18.41
FR_PAR: 42.10   -       2.50    72.50   94.06   72.20   166.26
US_WDC: 32.48   -       5.50    40.90   83.07   25.92   108.98
RO_BUC: 92.48   -       14.80   96.50   84.00   4.35    88.34
AU_MEL: 13.28   -       12.50   14.00   5.84    5.46    11.30
IT_MIL: 44.75   -       5.30    76.30   88.16   70.50   158.65
SG_SGP: 27.08   -       13.70   28.40   49.42   4.86    54.27
US_DAL: 18.50   -       3.10    33.40   83.24   80.54   163.78
RU_MOS: 55.09   -       3.90    69.30   92.92   25.79   118.71
JP_TOK: 14.53   -       1.60    18.80   88.99   29.36   118.35
NO_OSL: 31.12   -       2.30    65.10   92.61   109.16  201.77
IN_CHN: 12.87   -       2.10    20.40   83.68   58.55   142.23

Frankly, not a nice sight. Quite slow plus really high spans. Well, I guess there is only so much in terms of networking one can get in Moldavia.

Romania
GR_UNK: 83.67   -       69.10   87.60   17.42   4.69    22.11
US_SJC: 18.01   -       0.00    29.90   100.00  66.05   166.05
OK_LON: 47.63   -       5.90    82.80   87.61   73.86   161.47
DE_FRA: 44.57   -       3.20    82.80   92.82   85.77   178.59
BR_SAO: 24.50   -       22.20   26.00   9.38    6.14    15.51
FR_PAR: 48.48   -       3.10    82.20   93.61   69.56   163.17
US_WDC: 28.38   -       14.10   44.30   50.32   56.09   106.41
RO_BUC: 94.39   -       42.00   97.30   55.50   3.08    58.59
AU_MEL: 16.97   -       0.00    18.30   100.00  7.83    107.83
IT_MIL: 42.34   -       3.90    76.90   90.79   81.63   172.42
SG_SGP: 27.64   -       20.30   30.20   26.54   9.28    35.82
US_DAL: 20.60   -       1.80    35.40   91.26   71.87   163.13
RU_MOS: 73.28   -       63.30   79.20   13.62   8.08    21.70
JP_TOK: 14.19   -       2.40    21.00   83.09   47.95   131.05
NO_OSL: 36.70   -       2.60    72.10   92.91   96.48   189.39
IN_CHN: 27.21   -       23.30   30.10   14.38   10.61   24.99

Better, much better than the moldavian network, but still nothing to write home about, sorry.

Russia
GR_UNK: 51.51   -       19.70   57.30   61.75   11.25   73.00
US_SJC: 27.99   -       0.00    29.90   100.00  6.84    106.84
IT_MIL: 75.62   -       25.40   81.80   66.41   8.18    74.59
DE_FRA: 82.80   -       48.40   87.50   41.54   5.68    47.22
BR_SAO: 23.15   -       22.00   24.30   4.98    4.95    9.93
FR_PAR: 75.91   -       46.00   81.30   39.41   7.09    46.50
US_WDC: 36.43   -       19.20   43.90   47.29   20.51   67.81
UK_LON: 77.91   -       52.40   82.10   32.74   5.38    38.12
AU_MEL: 7.48    -       0.00    17.30   100.00  131.23  231.23
SG_SGP: 25.43   -       19.00   27.60   25.29   8.53    33.82
US_DAL: 33.82   -       30.60   37.00   9.51    9.41    18.93
RO_BUC: 64.88   -       35.40   72.80   45.44   12.21   57.65
RU_MOS: 94.78   -       48.90   97.80   48.41   3.19    51.59
JP_TOK: 18.80   -       0.00    21.90   100.00  16.47   116.47
NO_OSL: 66.46   -       44.30   70.80   33.35   6.52    39.87
IN_CHN: 26.25   -       23.20   28.20   11.63   7.41    19.04

Meh, not really stunning neither but something one can work with.

USA
GR_UNK: 7.21    -       1.50    21.20   79.21   193.83  273.04
US_SJC: 87.55   -       40.70   89.60   53.51   2.34    55.85
OK_LON: 12.82   -       3.40    35.70   73.48   178.51  251.98
DE_FRA: 10.66   -       3.10    28.70   70.93   169.10  240.04
BR_SAO: 10.32   -       3.90    21.20   62.22   105.36  167.58
FR_PAR: 11.93   -       4.20    34.30   64.81   187.39  252.20
US_WDC: 49.92   -       10.90   62.00   78.16   24.21   102.37
RO_BUC: 12.43   -       3.80    30.20   69.43   142.96  212.39
AU_MEL: 10.91   -       2.50    24.50   77.09   124.53  201.62
IT_MIL: 10.74   -       3.60    25.50   66.46   137.54  204.01
SG_SGP: 13.95   -       3.30    29.00   76.35   107.86  184.21
US_DAL: 72.65   -       44.10   77.70   39.30   6.95    46.25
RU_MOS: 13.65   -       4.10    28.90   69.97   111.67  181.64
JP_TOK: 39.14   -       8.80    56.20   77.52   43.59   121.10
NO_OSL: 10.92   -       3.60    27.70   67.04   153.62  220.66
IN_CHN: 8.83    -       3.00    20.40   66.02   131.07  197.09

Pardon me but this is disappointing. While the connectivity within the major US targets are kind of OK the rest is not. Plus terrible spans that make opening a connection a lottery game.

I don't like that because the mivo people were really friendly but considering that what I tested is sold as VPS3 (2 vCores, 2 GB mem. 50 GB disk) and costs €10 per month I doubt that many people want to become customers. Yes, they say that the KVM VPS comes with 1 TB DDOS protection (which one may believe or not ...), virtual firewalls, "unlimited traffic" (at 100Mb/s) but still in my book that's called expensive and not attractive. Sorry.

Thanked by 2uptime dahartigan

Comments

  • jsg said: costs €10 per month

    if I remember correctly, they also offer some service with hourly billing - but that is different from the monthly billed VPS instances you tested?

  • jsgjsg Member, Resident Benchmarker

    @uptime said:

    jsg said: costs €10 per month

    if I remember correctly, they also offer some service with hourly billing - but that is different from the monthly billed VPS instances you tested?

    Yes, it seems so.

    Thanked by 1uptime
  • jsg said:
    mivo people were really friendly
    [...]
    doubt that many people want

    some people pay extra for that (friendly service) :)

    point being that having a competent human in the loop is the one thing that is difficult to scale - at least at the levels we're typically dealing with on the low end. So - with regard to mivocloud (and similar providers being reviewed) - it might be interesting (if possible) to compare how they respond to feedback about their routing for a specific location. Some providers may be interested to know and willing/able to at least try to make improvements to help keep a customer happy. - while others will be quick to tell you that they just "don't have the bandwidth" to mess around with their routing just for you. And - depending on achievable results - that may or may not be worth an extra $7 per month or whatever.

    While not intending to belabor or even continue too much of the "reviewing the review" discussion from the other thread ... might nonetheless suggest a useful metric that goes beyond the comprehensive hardware/network analysis: that is, long-term reliability. While that's not exactly possible to assess in just a few days, it seems that you do make an effort to gauge some indicators perhaps even just in terms of (roughly speaking) ability to communicate effectively with a provider on a technical level (etc). Anyway, just thinking out loud here - thanks for another interesting review!

    Thanked by 1dahartigan
  • jsgjsg Member, Resident Benchmarker

    @uptime

    The scope of a review, even one that is much more than a quick shot, is limited. But over time this community itself provides some kind of a long term experience.

    As for friendliness, yes, that is certainly worth a lot - but how to judge, let alone measure that? I try in all my reviews to transmit something like my personal impression, but we all know well that the perception of friendliness is subjective to a large degree, so I'm not making a lot of noise about it.
    Another reason is the fact that users themselves are very different. A provider who is patient and friendly when asked technicalities might be quite dry and short cut, maybe even unfriendly with repeated questions he deems idiotic. Plus I guess that providers support people are often (ab?)used as teachers during someones first steps. Could I rightfully call some provider unfriendly if he instructs his support staff - whom he has to pay - to not help customers with their first steps in hosting?
    TL;DR - It's hard if not even impossible to offer more than a personal first impression in a review. But not all is lost; after all we all can experience most providers first hand here and can get an impression of them.

    Thanked by 1uptime
  • uptimeuptime Member
    edited November 2019

    jsg said: how to judge, let alone measure that?

    yes, that is a bit of a challenge. Though I'm not exactly suggesting that some intangible "friendliness" metric would even be directly relevant other than as a proxy for some more general wherewithal (plus the observation that people do sometimes pay extra for "friendliness"). And as you pointed out, that can also be a paradoxical indicator reflecting inefficient use of support staff (or perhaps in some cases inexperience and/or inebriation - but I digress...)

    More specific points relating to long-term viability might include years in business, "own hardware", ASN and so forth. I think you do often mention these sorts of things in passing, which is probably the most appropriate treatment in the absence of a rigorously applied formula and valid predictive model, etc etc etc ... :)

    And of course "past results are no guarantee of future performance" - maybe the best one might hope for is to identify obvious warning signs when apparent. (Which I would agree is one of the things that at least a few people paying attention on LET do tend to have a knack for!)

    Thanked by 1dahartigan
Sign In or Register to comment.