Subject: Re: [vserver] linux-vserver, lxc and docker, +++
From: Tor Rune Skoglund <trs@swi.no>
Date: Thu, 27 Mar 2014 10:33:17 +0100

Thanks, Herbert! :-)

Den 26. mars 2014 18:57, skrev Herbert Poetzl:
> On Wed, Mar 26, 2014 at 01:48:25PM +0100, Tor Rune Skoglund wrote:
>> I came across this presentation embedded on this page:
> 
>> http://bodenr.blogspot.no/2014/03/linux-containers-building-blocks.html
> 
>> On page 31, there is an overview of various virtualization
>> methods. What is community views on the comments made on
>> linux-vserver?
> 
> I'd say he didn't do his research properly.
> 
> First, we have been there long before OpenVZ, so the
> 'well seasoned' listed there is dubious.
> 
> Second, we are (now) based on cgroups and namespaces,
> so the summary there is wrong as well.
> 
> Further ...
>  - we have python bindings as well
>  - there are some (outdated) panels
>  - I would definitely consider it more stable than OVZ :)
> 
> But obviously it is biased (towards Docker) and just
> takes the fact that OVZ now is owned by Parallels as
> a 'sign of stability' :)
> 
>> It is/was my impression that current linux-vserver "builds" on
>> cgroups and namespaces features from lxc in the standard kernel
>> --- and also that new lxc implementation gradually replace
>> "custom" linux-vserver implementations?
> 
> Completely correct. Whenever some mainline feature gets
> useable, we try to incorporate it as soon as possible.
> 
>> Reason for me asking, is that we are building a system based on
>> linux-vserver, and likes to have some confirmation that we have
>> not chosen the wrong technology. In that context, a "merge" of
>> linux-vserver into improving lxc tools in the very long run, is
>> a very acceptable roadmap for us.
> 
> I don't think you've made the wrong choice, but obviously
> I'm biased as well :)
> 
> best,
> Herbert
> 
>> - Tor Rune Skoglund, trs@swi.no