Howdy, Stranger!

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


Benchmark results between contabo and hetzner and some questions for future purchasees
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.

Benchmark results between contabo and hetzner and some questions for future purchasees

lowendguy7lowendguy7 Member
edited March 2019 in Providers

Ok so I did a test I looked up online using openssl on both the servers. I had already used hetzner for many months and contabo I just bought today.

The hetzner package is the 2 core 4.90 euro one and the contabo is the 4 core pure ssd offer for 4.99 euro.

The command

openssl speed des des-ede3 dsa2048 hmac idea-cbc md5

Results...

HETZNER

The 'numbers' are in 1000s of bytes per second processed.
type             16 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes  16384 bytes
md5              93228.61k   199009.48k   360655.57k   443382.63k   474822.20k   486511.02k
hmac(md5)        38521.38k   113018.04k   281359.76k   423960.38k   511178.06k   523466.02k
des cbc          53739.98k    55034.23k    59615.72k    58957.06k    61409.86k    59363.04k
des ede3         21282.92k    21329.10k    19955.03k    21551.24k    20380.03k    20997.82k
idea cbc         72881.44k    78206.63k    74194.90k    74627.78k    75545.69k    77350.31k
                  sign    verify    sign/s verify/s
dsa 2048 bits 0.000424s 0.000308s   2359.6   3245.8

CONTABO

The 'numbers' are in 1000s of bytes per second processed.
type             16 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes  16384 bytes
md5              82391.63k   180280.25k   321132.24k   398101.06k   464116.53k   444220.10k
hmac(md5)        34503.41k   107159.46k   259284.16k   395273.65k   470774.24k   464486.40k
des cbc          49203.52k    50862.86k    51527.42k    54651.40k    52398.25k    50060.36k
des ede3         18621.20k    19999.63k    20464.61k    19212.19k    20046.80k    20446.79k
idea cbc         65532.16k    67078.64k    68161.26k    69449.76k    67557.62k    71162.10k
                  sign    verify    sign/s verify/s
dsa 2048 bits 0.000475s 0.000357s   2106.8   2800.3

Disappointing to see the results of contabo actually coming out worse with 4 cores which goes to show the core numbers mean shit then and just a sales ploy I guess. I knew people said the same but I didn't expect the results to be that bad in terms of being even worse than my last server.

I will test again a few times on contabo of course in case I caught a busy load time. I do still get an increase in hardware though since hetzner only gives 40gb whereas contabo gives 200gb (and ssd thought not really bothered about the latter). But cpu was a big factor for me wanting to change server (since encoding is the main task I want to do on it); HD being the other so I am going to have to have a think if I want to look again now.

With this info, is there any way to know what your speed would be like before having to purchase when using vps? With dedicated I imagine this isn't an issue since better hardware would mean better performance since you know it isn't being drained by others right?

How can I prevent wasting time buying and setting up only to find out the performance is worse in future when buying servers?

Is it more of a clear playing field when looking for dedis? and would that be better for my use case of encoding being my main task on the server? Encoding won't be 100% of the time it is probably 50/50 since I download videos, encode, then upload and do next batch etc. Well on second thought downloading/uploading only takes a small amount of time compared to encoding so yeah I guess encoding would be the majority.

EDITED RESULTS

HETZ
openssl speed -multi 4 des des-ede3 dsa2048 hmac idea-cbc md5

md5             158783.07k   396589.32k   683482.11k   833170.96k   889339.90k   892807.85k
hmac(md5)        71498.40k   211002.30k   497262.59k   765638.11k   937366.87k   989708.98k
des cbc         102615.39k   103834.43k   107148.63k   106851.33k   112970.41k   106318.61k
des ede3         40055.68k    40394.44k    41710.49k    42205.18k    41399.91k    41194.84k
idea cbc        141328.23k   142297.66k   147675.90k   153802.45k   146333.82k   134911.32k
                  sign    verify    sign/s verify/s
dsa 2048 bits 0.000257s 0.000184s   3896.2   5435.2

CONTABO
openssl speed -multi 4 des des-ede3 dsa2048 hmac idea-cbc md5

md5             318320.59k   684247.57k  1287913.81k  1606255.27k  1748724.39k  1780438.36k
hmac(md5)       145102.30k   406587.54k   927453.35k  1522987.01k  1841329.49k  1886999.89k
des cbc         206431.53k   207200.38k   206354.86k   201171.29k   201534.12k   210965.85k
des ede3         75670.68k    70532.35k    73114.03k    76198.23k    76038.14k    74716.50k
idea cbc        249697.41k   265957.82k   261571.75k   269843.11k   264407.72k   266480.30k
                  sign    verify    sign/s verify/s
dsa 2048 bits 0.000121s 0.000103s   8265.2   9721.7

That looks better!

HETZ
openssl speed -multi 2 des des-ede3 dsa2048 hmac idea-cbc md5

md5             160086.18k   376901.08k   658161.66k   844717.40k   973144.06k   915581.61k
hmac(md5)        66983.39k   216536.30k   507931.14k   759950.68k   932063.91k   966639.62k
des cbc          97949.62k   104717.91k   114193.41k   110012.42k   107959.64k   110291.63k
des ede3         40904.10k    40507.07k    43666.77k    43367.42k    43362.99k    43013.46k
idea cbc        143822.70k   152119.47k   149100.03k   151850.33k   150538.92k   150880.26k
                  sign    verify    sign/s verify/s
dsa 2048 bits 0.000253s 0.000215s   3955.4   4643.9


CONTABO
openssl speed -multi 2 des des-ede3 dsa2048 hmac idea-cbc md5

md5             148392.33k   348738.47k   653198.25k   849411.07k   858633.56k   858303.15k
hmac(md5)        69290.61k   220553.73k   515411.97k   776717.65k   914472.96k   934652.59k
des cbc          99582.55k   105304.06k   106903.89k   103675.90k   105816.06k   103230.12k
des ede3         39506.28k    38645.78k    37635.07k    38683.31k    39600.13k    39818.58k
idea cbc        131182.07k   135432.62k   136422.31k   139212.80k   140208.81k   140896.94k
                  sign    verify    sign/s verify/s
dsa 2048 bits 0.000276s 0.000189s   3620.6   5282.1

Why does hetzner outperform contabo on 2 thread?

This testing ended up being way more circuitous than just installing some other software.

Thanked by 1gks
«13

Comments

  • No one challenges hetzner. Hetzner is the definition of vps.

    I hope u didn't have to pay setup fee for contabo. I eventually cancelled due to network and performance issues.

    Thanked by 1eol
  • lowendguy7lowendguy7 Member
    edited March 2019

    yes I had to pay 4.99 for setup fee. Not going to break the bank though :) . Regardless it is good to get concrete info on the general talk around here of providers overselling etc. but I want to know how to avoid wasting my time in future.

    This makes me feel like vps is a smoke and mirrors advertising game. I would prefer to know from the outset what I am too expect in terms of performance, not to be lowballed once I already paid.

    So is it time to get into the dedicated field again?

    @cybertech said:
    No one challenges hetzner. Hetzner is the definition of vps.

    I hope u didn't have to pay setup fee for contabo. I eventually cancelled due to network and performance issues.

    Thanked by 1cybertech
  • deankdeank Member, Troll
    edited March 2019

    You are not going PMS for losing 4.99? Then you are not a lowendguy.

    There was a dude who wrote 2,000 words essay (complaint) over losing 24 cents during refund process with his former host.

  • gksgks Member

    I yet to try Contabo, I see them cheaper on the higher ram/cpus than netcup RS plans. But there is an issue with 400 mbits limit even for local network, I yet to clarify with them. Your test is good, thanks.

  • It's not the size that matters, it's what you do with it.

  • eoleol Member

    @DrCornFlakes said:
    It's not the size that matters, it's what you do with it.

    Lies.

  • Don't traumatize me if you please.

  • eoleol Member

    @DrCornFlakes said:
    Don't traumatize me if you please.

    Don't post lies if you please.

  • You always get what you pay for. Contabo is great for personal projects or even business if you know about their flaws. Hetzner on the other hand requires you to pay more but is premium too. Plus we have @Hetzner_OL Katie and her lovely co-worker (someone surely remembers the name?).

    Thanked by 1Hetzner_OL
  • I think you've only tested a single core/thread version of the benchmark (and that is of course also CPU frequency dependent). Did you try with the -multi 2 or -multi 4 option and see if the numbers change on Hetzner vs Contabo?

  • FalzoFalzo Member

    @lowendguy7 said:

    The command

    openssl speed des des-ede3 dsa2048 hmac idea-cbc md5

    Disappointing to see the results of contabo actually coming out worse with 4 cores which goes to show the core numbers mean shit then and just a sales ploy I guess.

    really???

  • FalzoFalzo Member
    edited March 2019

    @nullnothere said:
    I think you've only tested a single core/thread version of the benchmark (and that is of course also CPU frequency dependent). Did you try with the -multi 2 or -multi 4 option and see if the numbers change on Hetzner vs Contabo?

    beat me to it...

    on a sidenote: if there is no understanding of multithreading options at all, the encoding tasks won't work very well regardless the server OP is going to use...

  • I am not familiar with the test. Is it designed to be a multi-threaded test?

  • @poisson said: I am not familiar with the test. Is it designed to be a multi-threaded test?

    It is a (useful, but take it with a pinch of salt) benchmark - that's all. Without the multi option it's single threaded. Usually very useful to check on the HW accelerated ciphers like AES (and having the right flags in /proc/cpuinfo, i.e. cpu pass thru' for KVMs, is very useful to significantly improve performance for many crypto ops).

  • @Falzo said: if there is no understanding of multithreading options at all, the encoding tasks won't work very well regardless the server OP is going to use

    Hear hear. And just to add it's not uncommon (for me at least) to come across folks who expect magic to happen just because you have more cores when they've barely tuned anything to take advantage of those cores correctly. Many DB setups have other bottlenecks including IO and RAM before they start hitting serious CPU limits and yet I've found many clueless people jumping the gun onto more cores to make things faster.

    Sigh... I wish the right people had access to the right hardware.

    Thanked by 1Falzo
  • @nullnothere said:
    It is a (useful, but take it with a pinch of salt) benchmark - that's all. Without the multi option it's single threaded. Usually very useful to check on the HW accelerated ciphers like AES (and having the right flags in /proc/cpuinfo, i.e. cpu pass thru' for KVMs, is very useful to significantly improve performance for many crypto ops).

    Thanks for the explanation. I ask because I thought the OP might have made an erroneous conclusion about the number of cores. In a multi threaded setting, it is inconceivable that the contabo would underperform so shockingly. It looked like a single thread to single thread comparison and dissing the number of cores as "sales ploy" just felt like an unfair statement.

    Thanked by 1nullnothere
  • poisson said: It looked like a single thread to single thread comparison and dissing the number of cores as "sales ploy" just felt like an unfair statement.

    Exactly.

  • I am a bit sensitive to the multithread issue because I use R for data processing and the newest and fanciest processors mostly are a waste of money because the vast majority of the libraries are single-threaded. The perfect data science server as of now just require a super highly clocked single core with 256gb RAM lol.

    Thanked by 1vimalware
  • lowendguy7lowendguy7 Member
    edited March 2019

    @deank said:
    You are not going PMS for losing 4.99? Then you are not a lowendguy.

    There was a dude who wrote 2,000 words essay (complaint) over losing 24 cents during refund process with his former host.

    Lol! I certainly look after the pennies but not beyond reason where it would increase my blood pressure over such a small thing.

    Anyhow I can still increase my productivity quite a bit with the same kind of cpu as I had before but now with an extra 160gb to play with so I might stay the month although there is a 14 day money back I will need to check if it has setup fee included in that.

  • lowendguy7lowendguy7 Member
    edited March 2019

    @Falzo said:

    @lowendguy7 said:

    The command

    openssl speed des des-ede3 dsa2048 hmac idea-cbc md5

    Disappointing to see the results of contabo actually coming out worse with 4 cores which goes to show the core numbers mean shit then and just a sales ploy I guess.

    really???

    There is no need to be condescending. I said I never used that command before and I took the command from the article I read in good faith and it didn't mention anything about multi-core arguments.

  • @lowendguy7 said:
    There is no need to be condescending. I said I never used that command before; so I must never make any errors in order to live up to your high standards?

    That was the appropriate response. If you never used the command before, you would have posted your results and asked for plausible explanations before condescendingly claiming that the extra cores are a ploy.

    Those who live in glass houses should not throw stones.

    Thanked by 1Falzo
  • @poisson said:

    @lowendguy7 said:
    There is no need to be condescending. I said I never used that command before; so I must never make any errors in order to live up to your high standards?

    That was the appropriate response. If you never used the command before, you would have posted your results and asked for plausible explanations before condescendingly claiming that the extra cores are a ploy.

    Those who live in glass houses should not throw stones.

    ok i am worthless piece of shit. Happy?

  • @lowendguy7 said:
    ok i am worthless piece of shit. Happy?

    That's not the point. I am just saying that you are getting a comensuratble response. Also, it would be nice to put an edit that you changed the offensive tone in your last reply.

  • The single core is kind of disappointing tbh.

  • @poisson said:

    The single core is kind of disappointing tbh.

    Yes it was although I was more disappointed in multi core score and real time usage

  • FalzoFalzo Member

    @lowendguy7 said:

    There is no need to be condescending. I said I never used that command before and I took the command from the article I read in good faith and it didn't mention anything about multi-core arguments.

    Thanked by 1poisson
  • @cybertech said:

    @poisson said:

    The single core is kind of disappointing tbh.

    Yes it was although I was more disappointed in multi core score and real time usage

    I think for the price on an overall basis, it's pretty neat. It will work well as a RDP and you cannot get that core/RAM/SSD combo anywhere else. I was tempted to get one to out a gucamole on it.

    Thanked by 1cybertech
  • openssl -multi 2 speed des des-ede3 dsa2048 hmac idea-cbc md5
    Invalid command '-multi'; type "help" for a list
    
  • dude...
    openssl speed -multi 2 des des-ede3 dsa2048 hmac idea-cbc md5

    Thanked by 1lowendguy7
Sign In or Register to comment.