Howdy, Stranger!

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

Advertise on LowEndTalk.com
My experience with LeaseWeb vs. OVH (VPS SSD speed!)
New on LowEndTalk? Please read our 'Community Rules' by clicking on it in the right menu!

My experience with LeaseWeb vs. OVH (VPS SSD speed!)

Hi,

tl;dr:

LeaseWeb have much better random read/write performance on their SSD VPS, compared to OVH's, which is only better at sequential read and write.

I manage a small project that mostly targets visitors from Egypt, so I was looking for an entry level EU based VPS that have the fastest ping speeds from there. The choices I ended up with were OVH, Hetzner, and LeaseWeb.

I signed up with OVH, and used it for a couple of months, then decided to give LeaseWeb a try, since my OVH's VPS ping from Egypt wasn't the greatest, and LeaseWeb is a bit cheaper giving I don't need big amount of RAM.

So I went from:

OVH - VPS SSD 3 (2 vCores, 8GB RAM, 40GB SSD) located in their Strasbourg (FR) data center - €11.99/m

to:

LeaseWeb - VPS M (2 vCores, 2GB RAM, 60GB SSD) located in their German data center - €9.95/m

Both VPSs are running updated Ubuntu 16.04 with kernel 4.7

Apart from the vastly improved ping and download/upload speed (LeaseWeb provides 1Gbps interface vs. 100Mbps from OVH), I could clearly notice that compiling and setting up software was actually much faster on LeaseWeb's VPS! I decided to run some benchmarks to see if there is any actual difference in performance between both.

First I ran vpsbench

OVH:

Benching I/O ... OK
Benching CPU. Bzipping 25MB file ... OK
Benching inbound network. Downloading 100MB file ... OK
Share at https://github.com/mgutz/vpsbench/wiki/VPS-Hosts

08/20/2016 - VMPLAN - DATACENTER - OS - AUTHOR

CPU model: Intel Core Processor (Haswell)
Number of cores: 2
CPU frequency: 2394.470 MHz
Total amount of RAM: 7797 MB
Total amount of swap: MB
System uptime: 2 days, 22:43,
I/O speed: 317 MB/s
Bzip 25MB: 4.90s
Download 100MB file: 12.0MB/s

LeaseWeb:

Benching I/O ... OK
Benching CPU. Bzipping 25MB file ... OK
Benching inbound network. Downloading 100MB file ... OK
Share at https://github.com/mgutz/vpsbench/wiki/VPS-Hosts

_08/20/2016 - VMPLAN - DATACENTER - OS - AUTHOR

CPU model: QEMU Virtual CPU version 1.2.1
Number of cores: 2
CPU frequency: 2300.195 MHz
Total amount of RAM: 2002 MB
Total amount of swap: MB
System uptime: 19:30,
I/O speed: 84.1 MB/s
Bzip 25MB: 4.88s
Download 100MB file: 79.1MB/s

As you can see, network speed difference and Bzip test results are expected and normal... but WTH is wrong with LeaseWeb's VPS giving only 84MB/s on that SSD vs 317MB/s on OVH's!! This can't be right, as otherwise it should be much slower in compiling stuff and other IO intensive tasks. I tried the benchmark several times and each time I get similar results, so I decided to run another benchmark that will measure the IOPS. I used this method and here is the results of this command:

./fio --randrepeat=1 --ioengine=libaio --direct=1 --gtod_reduce=1 --name=test --filename=test --bs=4k --iodepth=64 --size=1G --readwrite=randrw --rwmixread=75

OVH:

test: (groupid=0, jobs=1): err= 0: pid=8485: Sat Aug 20 17:22:21 2016
read : io=786224KB, bw=6001.3KB/s, iops=1500 , runt=131015msec
write: io=262352KB, bw=2002.5KB/s, iops=500 , runt=131015msec
cpu : usr=1.11%, sys=5.09%, ctx=244975, majf=0, minf=4
IO depths : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0%
submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.1%, >=64=0.0%
issued : total=r=196556/w=65588/d=0, short=r=0/w=0/d=0

Run status group 0 (all jobs):
READ: io=786224KB, aggrb=6001KB/s, minb=6001KB/s, maxb=6001KB/s, mint=131015msec, maxt=131015msec
WRITE: io=262352KB, aggrb=2002KB/s, minb=2002KB/s, maxb=2002KB/s, mint=131015msec, maxt=131015msec

Disk stats (read/write):
vda: ios=196220/65583, merge=0/70, ticks=8302972/66400, in_queue=8370752, util=100.00%

LeaseWeb:

test: (groupid=0, jobs=1): err= 0: pid=22967: Sat Aug 20 17:23:15 2016
read : io=786692KB, bw=9368.5KB/s, iops=2342 , runt= 83973msec
write: io=261884KB, bw=3118.7KB/s, iops=779 , runt= 83973msec
cpu : usr=0.57%, sys=2.47%, ctx=60477, majf=0, minf=4
IO depths : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0%
submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.1%, >=64=0.0%
issued : total=r=196673/w=65471/d=0, short=r=0/w=0/d=0

Run status group 0 (all jobs):
READ: io=786692KB, aggrb=9368KB/s, minb=9368KB/s, maxb=9368KB/s, mint=83973msec, maxt=83973msec
WRITE: io=261884KB, aggrb=3118KB/s, minb=3118KB/s, maxb=3118KB/s, mint=83973msec, maxt=83973msec

Disk stats (read/write):
dm-3: ios=196647/65539, merge=0/0, ticks=4028688/1361600, in_queue=5390388, util=99.97%, aggrios=196673/65554, aggrmerge=0/57, aggrticks=4028676/1345132, aggrin_queue=5373748, aggrutil=99.91%
vda: ios=196673/65554, merge=0/57, ticks=4028676/1345132, in_queue=5373748, util=99.91%

So that was it! It looks like OVH have some cap on the IOPS, while LeaseWeb doesn't (or have a higher cap). And this translates to a very noticeable speed difference in favor of LeaseWeb's VPS.

I hope that was useful to anyone torn between the two in the said EU region. As long as you don't need a huge amount of RAM in that price segment, go with LeaseWeb.

Let me know if you need me to run any other benchmarks as I have the OVH VPS until the end of month.

Comments

  • joerijoeri Member, Host Rep, LIR

    Maybe you can run these tests

    wget https://www.janyksteenbeek.nl/linux/iotest.sh -O - -o /dev/null|bash

    wget https://www.janyksteenbeek.nl/linux/sysbench.sh -O - -o /dev/null|bash

    Then you will see the difference better

  • @joeri said:
    Maybe you can run these tests

    wget https://www.janyksteenbeek.nl/linux/iotest.sh -O - -o /dev/null|bash

    wget https://www.janyksteenbeek.nl/linux/sysbench.sh -O - -o /dev/null|bash

    Then you will see the difference better

    I did

    First one with OVH is in the 400MB/s range, with LeaseWeb is 90MB/s (which is similar to vpsbench result).

    Second test they are quite similar since it tests the CPU. BUT the installation of the needed packages was noticeably much faster on the LeaseWeb's VPS! (partly due to the faster network, and of course to the better IOPS)

  • JasperNLJasperNL Member
    edited August 2016

    Woah! Your story is very much the same as mine, especially regarding the part about Egyptians!

    2 years ago I used OVHs VPS (Classic 2014, not their SSD plan) with 1GB RAM. After running fine for a month I renewed the VPS for a year because their service was great, but very soon after that ther disk I/O started to go below 0.5 MB/s, their network was generating lots of package loss and the CPU started to be terrible. After two months of struggling with their support (tickets, phoning, mailing, forum posts) that didn't want to move me to another node or provide another alternative, I decided to take it as a loss and abandoned their service.

    That was the same period as when Leaseweb provided a 40% off coupon which encouraged me to make the move. I jumped on their smallest VPS plan and subscribed for two years. It was a great decision, as I've never had significant problems with them.

    A first small problem was that their storage platform for some days that caused offtime. As I'm not hosting anything business-critical, I was extremely satisfied with their 3 months free service because of the downtime for no more than 2 days!

    Another problem was with the Egyptian ISP TE Data, that decided to block an IP range from Leaseweb. Leaseweb told me they couldn't do anything about it, because it was due to the policy of TE Data. An Egyptian friend of mine decided to call that ISP and ask for an explanation, and the ISP told that they would be looking into my specific. A day later my IP was accessable from Egypt again. I don't know about the other IPs in the IP range, as that's not my concern.

    Lastly the bandwidth is something you should keep a close eye on. I didn't have problems with overage, because I'm using no more than 500GB/month while I pay for 4TB. It seems that you could be billed for an insane amount of money for receiving a DDoS attack, which frightens me a little. Also it seems that their measurement service is inaccurate.

    Summarizingly I'm very satisfied about Leasewebs service quality and support. Sometimes (mostly at impossible times, such as lunch breaks or during lectures) they call me to ask me about my satisfaction of their service and if they can help me with something else. This clearly points out that they do care about their customers.

    Oh, and I will never go with OVH again. I feel like they stole my money by putting noisy neighbors on my node just after I renewed for a year...

    Thanked by 1Tantawi
  • DouteiDoutei Member
    edited August 2016

    IIRC, LeaseWeb uses SAN for its vps lines, so the speed is limited by the network speed.

  • SmirSmir Member

    Wish LeaseWeb would do another promo, need something to get me away from OVH.

    Thanked by 1Lunar

    Smir

  • I only care about speed network not drive. ssd is useless if your site load very long times.

    lihat jadwal bola hari ini di tv online

  • Sixell said: IIRC, LeaseWeb uses SAN for its vps lines, so the speed is limited by the network speed.

    Correct

  • sinsin Member

    Here's a Geekbench of a Leaseweb Medium VPS (the $9.95/month one)
    http://browser.primatelabs.com/geekbench3/5349201 - I'm running a LEMP stack on it and everything runs great.

  • I really like LeaseWeb, it's a shame about their bandwidth overage
    policy: https://www.lowendtalk.com/discussion/74186/very-important-leaseweb-s-bandwidth-overage-policy

  • @madtbh said:
    I really like LeaseWeb, it's a shame about their bandwidth overage
    policy: https://www.lowendtalk.com/discussion/74186/very-important-leaseweb-s-bandwidth-overage-policy

    Doesn't limiting it to half your quota work? (even if their systems are slow in shutting off the port)

    Tired of LET scams?
    A Moderated forum : https://talk.lowendspirit.com/

  • jarjar Provider
    edited August 2016

    @JasperNL said:
    Another problem was with the Egyptian ISP TE Data, that decided to block an IP range from Leaseweb. Leaseweb told me they couldn't do anything about it, because it was due to the policy of TE Data. An Egyptian friend of mine decided to call that ISP and ask for an explanation, and the ISP told that they would be looking into my specific. A day later my IP was accessable from Egypt again. I don't know about the other IPs in the IP range, as that's not my concern.

    By the way, this is totally a thing right now. Egypt is blocking encrypted traffic (and more) to IP blocks at random. Protest it loudly. I don't know why it's not a huge international story.

  • @jarland said:
    I don't know why it's not a huge international story.

    Because Egypt is an ally and they dont want to complicate the good guy / bad guy story?

    Thanked by 2jar aboanas93
  • @vimalware said:

    @madtbh said:
    I really like LeaseWeb, it's a shame about their bandwidth overage
    policy: https://www.lowendtalk.com/discussion/74186/very-important-leaseweb-s-bandwidth-overage-policy

    Doesn't limiting it to half your quota work? (even if their systems are slow in shutting off the port)

    Depends on how quickly the bandwidth gets used up. It could be possible for you to use the entire allocation of bandwidth before the cap gets checked every 24 hours.

    They state in the control panel:

    Kindly note that the below measurements are done once in 24 hrs and are reset on monthly basis.

    Thanked by 1vimalware
  • madtbh said: They state in the control panel: "Kindly note that the below measurements are done once in 24 hrs and are reset on monthly basis."

    You're right :O I don't remember that it was there when I set it ages ago!

Sign In or Register to comment.