[Atlas-anchors-pilot] My answers to the questions from the presentation at RIPE65

[Romeo Zwart]

Hi Konstantin,

On 12/09/28 08:25 , Konstantin V Bekreyev wrote:
> 
> Hello,
> 
>> Do you want to be a host? Let us know!
> 
> Yes, this project is interesting to me.

That is good to hear. :)

>> HW Version A or Version B:
>> • Version A: 3.5” disk, 3 year contract, 3200.00 Euro
>> • Version B: 2.5” disk, 4 year contract, 3550.00 Euro
> 
> I would like to first try on our virtual machine VMware, and only then acquire server. Virtualization still will not get attached to any particular hardware, so it is profitable now.

To clarify the project goals: we are considering various virtualisation
options to allow us to run multiple RIPE NCC services on a single
system, as a next step beyond the Atlas Anchor deployment. For various
reasons (technical, procedural, maybe even legal) we currently do not
see it as an option to run Anchor code on a virtual system on iron
controlled by a third party.

In other words, currently we do not see your suggestion as a viable
option for the Anchors. It's not impossible that we would investigate
this as an alternative at some point, but at the moment that is really
far beyond the horizon.

> If we buy a server with a reserve for the future, of course, is more profitable to buy a server with 2.5" drives, because the 3.5" may soon cease to exist.

>> Virtualization issues:
>> • what kind of measurements would you want to do, that virtualization would not be helpful for?
> 
> I do not see any problems with virtualization, only advantages.
> 
>> • do you value “multiple services on the same box”, even if the “price” is to have to use virtualzation?
> 
> Of course, with virtualization resources are used more efficiently, and it reduces the cost of power, cooling and maintenance. For example, we can, if necessary, to throw a virtual server to another hardware, is transparent to other participants. We can also increase the capacity of the server as needed, which will reduce the initial cost.
>