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.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
What is the future of OpenVZ?
As Ubuntu is releasing the next LTS 16.04, Linux kernel is moving towards 4.x. OpenVZ is still staying at 2.6.x. OpenVZ seems to still hold a firm position in the VPS market. So will OpenVZ be replaced by something new like LXC? Or will it follow the newer kernels? Apparently OpenVZ already missed the whole 3.x kernels. Your thoughts?
Comments
funerrary openvz the next generation ?
The kernel version numbers are meaningless - Linus decides arbitrarily when to move from 2 to 3 to 4 etc.
The "next version" of OpenVZ is based on the 3.x kernel and is available for testing. The latest and greatest is frequently not the greatest on servers. New and shiny isn't as important as proven.
Still, the future of OpenVZ doesn't look specially bright, long-term.
Why do you say that? If you look at the next-gen, it looks better than ever.
Always lagging behind, always needing to patch/workaround a lot of things... there is certainly a market for OpenVZ, but I think it will be less commercially viable over time.
Even the increased cost of IPv4 will be a factor, since cheap containers will not be as viable as before.
Even more overselling
It has its place. Sometimes stability is better than cutting edge features...
Are you referring to the kernel? What is an example of a real-world consequence that affects a wide range of people using OpenVZ?
always needing to patch/workaround a lot of things...
Define "a lot"? I"m not sure what an end-user would need to "patch"? There are some things OpenVZ cannot do, but that is and always has been, an inherent part of the design. Like anything, if it doesn't do what you need it to do, it's the wrong product to be using.
I don't see any indication of that myself, but quite the opposite.
OpenVZ wasn't created to foster cheap containers, but you could argue it's existence is partially responsible for creating the concept. Humans decide what OpenVZ is used for. If you leave the microcosm of "cheap containers", it has a wide variety of uses.
I know that OpenVZ wasn't created to sell cheap containers, that's why I specified "commercially viable". Still, for applications different than selling to end users, LXC and others are creating great competition.
If cheap containers can't be a thing any longer, the OVZ VPS market is going to shit long term. Nearly no one would sell OVZ or VServer if they can afford KVM or VMware.
But it doesn't really have that, at least not recently =\ We've had more crashes in anything > stab100 than we ever had in the year+ that we were running a ~stab90 kernel throughout the company.
If you've been following their stable kernel releases you'd see that either:
Personally, we're doing our new KVM plans with the goal of pulling users off OpenVZ just because it makes up a fair bit of ticket overhead ("OMG I can't modprobe", "why can't I use ipset?", "Why can't I use any of these iptables examples I found on the internet?", "where's MASQURADE support??").
Francisco
But, how come providers like ramnode offer rock solid ovz vps hosting? Some times I feel, their openvz is far better than kvm/xen offered by my other providers!
OpenVZ 7 is comming later
I'm not say ovz is crashing all the time, we have had most of our nodes with over a year of uptime and @mitgib had at least one that was like 2 years.
The problem is that Parallels is trying to find a way to monetize openvz now and it's going to end up getting messy since they're trying to merge things from virtuozzo that I'm strongly against.
Francisco
OpenVZ is still a robust system-level virtualization, there are a lot of differences with KVM and if they keep updating it (actually it's too outdated), it won't die.
Of course full virtualization is better for some things, but there is still need for something with very low overhead, and i think such need will be there in the future too. The true competitor for OpenVZ is not KVM, but LXC. However for now i don't see LXC killing OpenVZ.
See how Xen quietly died off in favour of KVM. Same will happen with OVZ, maybe only a few bottom of the barrel overselling providers will keep lumping along on some ancient over-patched with nasty hacks but still crashing kernel, the rest will move to KVM.
LXC comes standard on Ubuntu 16.04 Server now
LXC has some big issues, the biggest I spotted was no way to limit diskspace as well as no way to pass through 2nd level quota support. You can hack it together using an LVM to limit the users space, but they still can't manage quotas.
Francisco
No one have an opinion on Virtuozzo?
KVM is not a replacement, the market simply does not work that way.
OpenVZ will do fine until there is a viable alternative, like KVM destroyed Xen in this market segment, you cant oversell KVM by 16x and keep it running so its not viable yet.
If something like LXC/D does get picked up by a commercial hosting panel then maybe, but it has a long way to go.
Essentially it has no real competitor yet, but that advantage wont last forever.
It's expensive & clunky. It has/had a nice UI back in the day though.
Francisco
Anybody has tested Virtuozzo 7?
https://download.openvz.org/virtuozzo/releases/7.0-beta3/x86_64/iso/
Clunky how? I'm not very well versed in container technology but virtuozzo seems pretty solid. To me it seems to function more like an actual server than with OpenVZ. I've always had issues, particularly with networking, with OpenVZ. Virtuozzo seems to be easier, for me at least, to adapt to. But I haven't touched OpenVZ since 2010...
Lol...people were saying that 5 years ago too. Just because they don't use the lastest shiny object kernel number. Silly humans.
The next version of OpenVZ is called Virtuozzo and supports both OpenVZishness and KVM and can work with the virtuozzo tools and the virt tools. It incorporates a lot of LXC features as well. Almost everything it needs in kernel space is now in mainline 3.x kernel. I think their goal is to use 100% stock mainline kernel although not sure if they are quite there yet. I believe they have their own people with voting rights on the Linux Kernel working group.
Virtuozzo is now in public beta and scheduled to go stable this summer I think. It will probably be another year after that before most established hosters will be comfortable enough with it to switch en mass.
OpenVZ now called Virtuozzo future looks extremely bright imho but only time will tell. There is nothing else out there that comes close to the hardware efficiency and low overhead of OpenVZ/Virtuozzo and that is the bottom line that most decisions are based on at the end of the day.
Whatever it is Ubuntu is doing with LXC is not that much different from what I have read. Ubuntu just wants to be different as usual because that is how they think they can get more market share. Their motivations are usually not purely technical.
Oh really? Maybe you need to fire you sysadmin...you?
top - 12:54:43 up 264 days
top - 12:56:14 up 235 days
top - 12:56:48 up 265 days
top - 12:58:09 up 251 days
2.6.32-042stab108.8
And the only reason it's not a lot higher is because of the kernel updates (I don't used kernelcare) or hardware changes etc.
There's multiple hosts complaining about the 100+ series. I've got some in the 108 branch which seems fine for now but as I said the 113 branch is a mess, you can confirm that yourself if you check the release logs.
Francisco
So what do you suggest a sysadmin could do when user-ran software locks the entire node? Because it happens.
Edit: I was referring to reboots, not kernel issues with that particular set of releases. Not sure if I misread or you edited the post.
You also forgot to mention how serious were some of those vulns which required a reboot
Oh Really? Tell me more.