CPU Priority

PPW Admin

KH Member since 2008
On the new VPS packages, what does "CPU Priority" mean?

Example VS-5 says "4x Priority" but VS-6 says "8x Priority".

What was the CPU Priority settings on the old Hybrid plans for reference? (HY-1)

Would either of these new VPS plans be more powerful (CPU access, load handling) than the previous HY-1 plan?
 
With old plans the same priority was setup for all hosting plans, now higher hosting plans will receive more CPU time in the event of 100% CPU usage. For example, if VPS-1 (base priority) and VPS-5 (4x) will attempt to consume 100% of CPU time at the very same moment VPS-5 will receive 4 times more CPU power than the VPS-1 one.
 
With old plans the same priority was setup for all hosting plans, now higher hosting plans will receive more CPU time in the event of 100% CPU usage. For example, if VPS-1 (base priority) and VPS-5 (4x) will attempt to consume 100% of CPU time at the very same moment VPS-5 will receive 4 times more CPU power than the VPS-1 one.
Ok, that makes sense.

Will VPS plans be mixed on the same node? (i.e. Node 1 is only for VPS-1 plans, node 2 only for VPS-2 plans, etc?) If they are split, how many VPS's will be on a VPS-6 node?

(Hope that question makes sense).

Basically I'm looking for a plan that is at least as powerful (prefer a bit more powerful) in CPU/Load handling/Node density than my current HY-1 plan.
 
We'll have a mix of different VPS plans on every machine.

VPS-6/VPS-7 plans were developed as a direct replacement for old Hybrids. With the new VPS product line we're not going to create "hot spots" with multiple resource-intensive systems running on the same hardware which is exactly what old Hybrid product line was but would rather spread them across many underloaded servers to improve overall resource availability and usage.
 
I have been wondering the same thing as the OP as my main concern is that moving from HY-1 to VPS-6 might result in system degradation.

As it is, since the announcement I've seen multiple high server load messages, some when there are a minimum of users on the website, so I'm assuming either there have been other clients moved to the same node/server to rationalise the number of nodes/servers or there is a larger number of moves going on than normal. As another example, a maintenance script that normally takes 2 or 3 minutes to run took 27 minutes last night.
 
Aussie_Boy,

We haven't started re-balancing / migration process yet. We plan to start working on this sometimes in January or February depending on work load, etc. As of right no (old)Hybrids were relocated anywhere and no new accounts were added to the (old)Hybrid-only servers since the moment when new line of hosting plans was introduced. So, unfortunately, the assumption you've made is not correct.

You're move than welcome to keep your HY1 hosting plan if you'd like. We are not forcing anyone to move to the new product line. If you don't feel comfortable with new plans you can continue to use the old one up until the moment when you'll decide to switch to one of the new plans.

Regards,
Paul
 
Aussie_Boy,

We haven't started re-balancing / migration process yet. We plan to start working on this sometimes in January or February depending on work load, etc. As of right no (old)Hybrids were relocated anywhere and no new accounts were added to the (old)Hybrid-only servers since the moment when new line of hosting plans was introduced. So, unfortunately, the assumption you've made is not correct.

You're move than welcome to keep your HY1 hosting plan if you'd like. We are not forcing anyone to move to the new product line. If you don't feel comfortable with new plans you can continue to use the old one up until the moment when you'll decide to switch to one of the new plans.

Regards,
Paul


Thanks Paul.

Is your gut feeling that the new VPS-6 will work at least as well as the HY-1 for an active MySQL based forum?

If there is obvious system degradation, is it possible to move back?
 
Is your gut feeling that the new VPS-6 will work at least as well as the HY-1 for an active MySQL based forum?

Well... We will not be releasing a new product line if we'll feel that overall quality of our service would go down. The rest depends on specific conditions, specific load pattern, etc.

If there is obvious system degradation, is it possible to move back?

No, there is no way back.
 
I upgraded my HY-1 to VPS-6 yesterday mainly due the the "CPU Priority 8x" that is offered. I run a very busy, MYSQL intensive forum on my server. So far performance seems the same, but as Paul has stated, they haven't completed any load balancing yet, so I will expect performance to improve once the nodes are balanced out.

(I have had many instances of unexplained load spikes in the past when my site was at a slow period, I am guessing that other Hybrids on my node were hogging resources. I am hoping the new CPU Priority as well as the coming load balancing will greatly improve the performance of my VPS going forward, time will tell.)
 
Have had 2 instances of Hyperspin noting my server as "Down" since moving to the VPS-6, I had this occur occasionally on my Hybrid as well. These typically happen in the middle of the night when my server/site is very slow so I am assuming it is another Hybrid on my node hogging resources.

When will the re-balancing of the nodes begin? Also, we EVERYONE on a given node have some type of "CPU Priority" setting so that older VPSs/Hybrids that don't have this setting can't hog all of the resources? It would seem mixing "old and new" would lead to a problem of some having the CPU Priority setting (and respecting that limit) while others wouldn't, which wouldn't be fair.

Thoughts?
 
Could you please either create a ticket with downtime timestamps and ask that ticket to be escalated to me or PM me the timestamps info so I can investigate this for you.
CPU allocation priority with most of the new VPS plans is higher than the old ones. Also, majority of our servers show at least some CPU time in the idle state pretty much all the time. We do a pretty good job finding and "firing" CPU and disk I/O abuses...
 
Hi Paul,

Ticket 272910 on 11/25
Ticket 277546 on 12/16
Ticket 279437 on 12/24
Ticket 280640 on 12/28
Ticket 282123 on 01/04

These are the 5 most recent "down times" as reported by Hyperspin.

Thanks for your help!
 
Hi Paul,

Any update on the status of the node re-balancing? I would imagine the initial surge of customers changing plans from the Christmas update is behind us and the node balancing would being soon (or perhaps has already begun)?
 
PPW Admin - yes, we've started working on (re)balancing last week.
Excellent! Thanks for the update, is there an estimate on when the project will be finished so we can determine our "final" node and try to see if we have any performance improvements?
 
PPW Admin,

This project will never finish, this is an ongoing effort that is running pretty much non-stop for 5+ years. This work was paused for about 3 weeks after last Christmas to accumulate information associated with the new plans.
 
PPW Admin,

This project will never finish, this is an ongoing effort that is running pretty much non-stop for 5+ years. This work was paused for about 3 weeks after last Christmas to accumulate information associated with the new plans.
I understand, it just sounded like with the new plan structure there would be a large restructuring of the nodes to evenly spread out the types of VPSs. From what I understand, prior to the new plans Hybrids were on their own nodes, etc. I would think with the new CPU priority settings that you would now want a node with a mix of VPS1- VPS7 accounts to spread the load more evenly.

I guess my thought was that after the "large restructuring" was done, a notice would be given to let us know that things were now "set" in the new layout (obviously, on going balancing would still occur as it has in the past).

As always, thanks for the responses and great support! :)
 
Top