Howdy, Stranger!

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


OpenVZ is pointless
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.

OpenVZ is pointless

What's the point in purchasing anything other than dedicated (KVM instead of OpenVZ)?

When you needed it, it wouldn't be there.

It's basically imaginary. It'd be better to buy a smaller KVM.

OpenVZ as a whole is a really flawed idea for anything semi production, since it's not remotely dedicated and barely isolated with a shared kernal.

Personally, I think OpenVZ should be put to death. Most "Cloud" hosts now only will use Xen or KVM.

The only positive of OpenVZ is it needs slightly less resources because of shared resources, but it's an old technology.

Discuss your opinion below.

«1

Comments

  • I usually get 2GB/s write speed on my NVMe OVZ VPSs.

  • OpenVZ is much more efficient. It is good for most applications, and even better than KVM for some.

    It's a shame that Virtuozzo 7 still isn't production-ready.

  • Well, I like cheap openvz vps that are $5-10/y. You can't get that with kvm virt. I don't care in fact, it's mean for idle anyway ;)

  • Crandolph said: It'd be better to buy a smaller KVM

    Such a thing is generally not being offered. There's not much choice in low end VPS. Did you mean LARGER KVM?

  • OVZ scales better and tends to be cheaper, no more a security risk than hosting with a random KVM host you can't vouch for.

  • @Ole_Juul said:

    Crandolph said: It'd be better to buy a smaller KVM

    Such a thing is generally not being offered. There's not much choice in low end VPS. Did you mean LARGER KVM?

    Depends on your definition of small. 256MB for me is small.

    Obviously you'd get a smaller KVM in an equal comparison to OpenVZ, pricing wise. Because of overselling of course.

  • @ricardo said:
    OVZ scales better and tends to be cheaper, no more a security risk than hosting with a random KVM host you can't vouch for.

    Don't use a random one then. Linode, Digitalocean, Vultr, etc, all not bad choices with pros and cons. At least companies that won't disappear overnight.

  • @FredQc said:
    Well, I like cheap openvz vps that are $5-10/y. You can't get that with kvm virt. I don't care in fact, it's mean for idle anyway ;)

    For super small personal projects like an OpenVPN server for example I would say yes OpenVZ is ideal in that narrow scope of things (or the idling project, that too).

  • @bugrakoc said:
    OpenVZ is much more efficient. It is good for most applications, and even better than KVM for some.

    It's a shame that Virtuozzo 7 still isn't production-ready.

    Yeah this is also an issue. Virtuozzo 7 isn't ready and OpenVZ is super old.

  • @lion said:
    I usually get 2GB/s write speed on my NVMe OVZ VPSs.

    dead meme

  • Crandolph said: Don't use a random one then. Linode, Digitalocean, Vultr, etc, all not bad choices with pros and cons. At least companies that won't disappear overnight.

    You missed the context of cheaper

  • wwabbitwwabbit Member
    edited February 2018

    When I own the node and all the virtual servers that are to be created on it, then KVM is pointless and a waste of resources compared to any container based visualization.

  • It's a better investment compared to giving the Prince of Nigerian my personal information to become the richest man alive.

  • @wwabbit said:
    When I own the node and all the virtual servers that are to be created on it, then KVM is pointless and a waste of resources compared to any container based visualization.

    Yeah but then you also have dedicated resources. So in this case, OpenVZ could work since you know if your overselling it or not.

    Regardless, I'd prefer KVM.

  • hostdarehostdare Member, Patron Provider

    We also starting to offer KVM only these days except some legacy plans or if we feel that we need to give some promo ovz plans .
    At this point KVM is only future unless something changes drastic over openvz 7 .

  • Been using OpenVZ as production for several years

    Thanked by 2jvnadr gestiondbi
  • hostdarehostdare Member, Patron Provider

    @jcaleb said:
    Been using OpenVZ as production for several years

    I think next year is eol ,time to move ;)

  • hostdare said: I think next year is eol ,time to move ;)

    What do you think will happen to the very many OpenVZ instances currently being operated around the globe?

  • @Ole_Juul said:

    hostdare said: I think next year is eol ,time to move ;)

    What do you think will happen to the very many OpenVZ instances currently being operated around the globe?

    They will immediately, suddenly, without warning or whatsoever just stop working and all IPs will resolve to a cat.jpg

    It’s in the docs. Somewhere.

  • LeviLevi Member
    edited February 2018

    :D euking haters... OpenVZ will live long and prosper among those who seek cheap services. Do you sleep better when you know that your server is kvm? Fuck kvm, go LXC.

  • FranciscoFrancisco Top Host, Host Rep, Veteran
    edited February 2018

    @LTniger said:
    :D euking haters... OpenVZ will live long and prosper among those who seek cheap services. Do you sleep better when you know that your server is kvm? Fuck kvm, go LXC.

    Quit with that, no large scale host is going to use LXC for the following reasons:

    • Load Averages. Load averages are not calculated on a per container basis like OpenVZ. This means that a user with a completely idle container is going to see the full load of the node and more likely than not will complain to the host or even cancel because "the VPS is broken/node is oversold".

    • Storage. Disk Quota's are not easy to implement. Since there is no simfs or ploop, hosts have to use LVM (which means they can't oversell their space at all), BTFS (which means they don't give a single shit about customer data), or ZFS (which works mostly fine but has problems documented in the following point). Sub quotas (so the user can set their own users space) only works on LVM and requires a lot of patching around to make it supposedly play nice. Now that you're on LVM shrinking a users volume is going to be iffy and a much greater chance of data loss or other issues. You also can't bump a users inodes without also bumping their total disk allocation. If you formatted the drive to allow many inodes then that's simple enough, but still.

    • Security. OpenVZ was designed from the ground up to be multi-tenant and with the mentality that 'we cant trust the user and we need to do our best to protect everything'. LXC was designed more so for a user to containerize their own applications, not to be leasing things out. It is extremely easy to make a container have complete full root access to a node. You're more or less required to use something like AppArmor so that you can blacklist a lot of /sys and things like that, otherwise you end up leaking a ton of data about your storage, configuration settings, etc.

    • No IP locks. LXC doesn't have VENET or anything similar to it, meaning a user is getting a fully bridged interface and by default will be able to ARP, bind, etc, any IP address they want. You'll be forced to put your own protection in here to hopefully curb it.

    I'm sure I could think of a bunch more problems most hosts will run into using it.

    Francisco

  • If I go full stupid... docker image based vps? :{
    https://store.docker.com/images/debian

  • hostdarehostdare Member, Patron Provider

    @Ole_Juul said:

    What do you think will happen to the very many OpenVZ instances currently being operated around the globe?

    They will keep running as customer keep paying for it .security will be more major concern .

  • @Francisco said: Quit with that, no large scale host is going to use LXC for the following reasons:

    Not sure why you took @LTniger so seriously, but your comment was useful nevertheless. :-)

  • @angstrom said:

    @Francisco said: Quit with that, no large scale host is going to use LXC for the following reasons:

    Not sure why you took @LTniger so seriously, but your comment was useful nevertheless. :-)

    As the next summerhost™ might go with LXC & cause an even bigger shitshow.

  • randvegetarandvegeta Member, Host Rep

    What's wrong with Virtuozzo 7 exactly? Send it's been in production for quite some time already?

  • @Aidan said:

    @angstrom said:

    @Francisco said: Quit with that, no large scale host is going to use LXC for the following reasons:

    Not sure why you took @LTniger so seriously, but your comment was useful nevertheless. :-)

    As the next summerhost™ might go with LXC & cause an even bigger shitshow.

    I think that there are a few established hosts on LET who offer LXC (e.g., I recall seeing at least one offer in the past by @cociu for LXC), but yes, indeed, it's not (yet) an appropriate replacement for OpenVZ.

  • Ovz will be around for a long time. I am sure about that because there are people using linux versions with ancient kernels and windows xp.

  • FranciscoFrancisco Top Host, Host Rep, Veteran

    angstrom said: Not sure why you took @LTniger so seriously, but your comment was useful nevertheless. :-)

    It doesn't matter if he's trolling or whatever, there's just this growing idea that LXC is this perfect choice to upgrade OpenVZ into.

    Many hosts on here are operating on razor thin margins or no margins and are hoping to upsell down the road. If these hosts suddenly have to go through all this extra work making sure things aren't going to catch fire or having to deal with a mountain of explanation tickets, they're going to get vortex'd.

    We've all seen countless hosts that (as an example) open a thread in February asking for help installing CentOS on a server, and by late March they're posting offer threads.

    At least with OpenVZ there's some security in the platform.

    I've had way too many tickets where a users trying to talk me into giving them extra permissions they should never have. I at least know better but newer hosts may very well give that access. Mix that with LXC being insecure by default and it's a gongshow.

    OpenVZ 6 will continue on for a long long time. The kernels have been in maintenance-only mode since OVZ7 got announced but if someone like KernelCare decides to keep backporting whatever exploits come to it, it'll stay alive and well.

    I simply don't see many hosts putting any effort into it. I don't think any host with their own in-house panel is going to invest any time into OVZ7 either. OVZ7 as a whole just feels like it's Virtuozzo-lite. They aren't supporting debian users at all, their templating engine is garbage, development also seems to be on the fritz.

    I don't know what happened with OVZ as a whole. Once the 2.6.32-100+ series came about I had mountains of crashes and they had no less than 3 - 4 filesystem breakouts, many of them were the same exploit popping up again because of poorly handled merges.

    The 90 series were super stable. I had boxes with well over a year uptime on the 93.5 ones. Solid kernels.

    Francisco

  • @Francisco said: angstrom said: Not sure why you took @LTniger so seriously, but your comment was useful nevertheless. :-)

    It doesn't matter if he's trolling or whatever, there's just this growing idea that LXC is this perfect choice to upgrade OpenVZ into.

    Do you think that LXC has the potential to replace OpenVZ down the road? (Assuming that the security issues can be addressed appropriately.) Or would this be unrealistic in the next few years?

Sign In or Register to comment.