CPU Priority and Guaranteed RAM as stipulated in each vps package......

Alpha

New Member
Hi,

Even though it has been over a year now with KH and being a non-techie, I am interested to know and better understanding of the following :
1. what does it mean by having 2x, 4x and 8x priority for different vps packages?
2. what is the advantage of guaranteed RAM to a vps account such as in my case - vps3 with the current 2560mb guaranteed RAM?

Thank you in advance.
 

Dan

Moderator
Hi Alpha,

1. what does it mean by having 2x, 4x and 8x priority for different vps packages?
I don't know the actual specifics on this but I would say that it simply means that the higher the number the higher your priority to use the CPUs available in the host machine. If I'm wrong about that I would love to hear what it really is :)

2. what is the advantage of guaranteed RAM to a vps account such as in my case - vps3 with the current 2560mb guaranteed RAM?
Guaranteed ram is offered as opposed to burstable ram. Previously (and still offered at many other places) there would be a smaller amount of ram offered with a large number of burstable ram. What that means is that when your system requires more ram than it's assigned it can 'burst' up to said amount but would then have to release that memory back to the system. I've actually seen hosts who would drop you if you used that burstable ram too frequently or for too long. And of course they never tell you what is 'too' often or 'too' long ;)

Hope that helps!
 

KH-Jonathan

Director of Managed Services
Staff member
@Dan

Great explanation on RAM.

@Alpha

Here's an explanation I've previously given to detail CPU priority:

Jonathan said:
CPU Priority (In OVZ/VZ) ONLY matters IF the node's CPU is 100% utilized. At this point, the containers with a higher priority do indeed get more CPU time.

It's all based on a ratio system. If you have a total of 10x priority, one container with 5x, 2 with 2x, and 1 with 1x - they'll each get their share based on the ratio of the priorities. The 5x will get 5/10 or 1/2, the 2x will get 2/10 or 1/5, and the 1x will get 1/10 of the total CPU power.

The core system is a bit differnet. It applies at all times and says you can use no more than X CPU - in this case one core.

The priority system will give you much greater flexibility to handle spikes over a core-based allocation assuming the node is managed properly and has available CPU - as in theory on a properly managed node that's not ever at max the CPU priority will have no effect and you could theoretically use all of the available CPU on the node - at which point it would hit 100% and priorities would knock you back down - but assuming you didn't make the node hit 100% you'd simply have more CPU available, your requests would process faster, thus the load wouldn't hit 100% as things wouldn't be queued up because since you had access to the extra CPU your spike was no big deal
awww.webhostingtalk.com_images_wht_smilies_smile.gif
 
Top