% CPU Usage (Demand) value in Cluster Capacity Risk Widget in vC Ops vApp 5.8.1

[Mauritz84]

Hi,

We are trying to understand a value in the detals table for the Cluster Capacity Risk widget. % CPU Usage (Demand) is the one which gives us in the table two values i.e. 37%/(72%). I can understand the value without the curly brackets but not the one inside them. Is this a cpu demand value of some sort, is it a derived metric vC Ops calculates for us?

What does it tell us and how should I interpret it?

For some clusters we see double for the value inside the curly brackets than % cpu usage and no stress but with redmarked % CPU allocation with really high numbers (above 700 %).

The picture is not taken from our environment. 

If anyone has any knowledge about this and maybe a source where to find more information, it would be much apprecierad.

Cheers,

Erik Alm

[mark_j]

This view is influenced by your display settings' non-trend view settings (interval type and # buckets). Basically, what size buckets of time and how many bucket to look back to evaluate in coming up with this data. Changing this will noticeably influence your views.

In looking at the view, they appear to be as follows:

% Mem Usage= Memory|Usage / Usable (%)

% Mem Demand = Planning|Usage Percent Trend|Memory|Demand

% CPU Usage= CPU Usage|Capacity Usage (%)

% CPU Demand= Planning|Usage Percent Trend|CPU|Demand

So, 'what your using now' / 'what you're trending towards actually demanding (growth)' for these couple fields in that view.

So if you're growing your workload, the trends will start going upward and you'll start seeing your demand trend getting to be higher values.. slow down growth and the trend will dip lower.

[Mauritz84]

Hi Mark,

Thanks for your detailed answer.

If you could help me I have some follow-up questions regarding these values. The have demand configured for CPU in the policy.

Our customers are seeing high values for % CPU Usage (43%) and Demand, (94%). The capacity remaining for the cluster is 5.5% or 19 VMs but with time remaining > 1 year.

The workload for the cluster is only 23 and I can't find any significant higher costop and ready.

The customer have been undergoing large migrations during the last months so they've decided to se their interval to Daily and 1 with 10 buckets. But even if I change this to quarterly with 4 intervals and 20 intervals, but it's still high.

The demand in the operations tab for thius cluster is 23% of configured well within the normal demand and only slightly higher than usage.

What could be the reason to these high values. Are they trending towards higher demand because of the capacity remaining?

Thank you!

Erik Alm

[mark_j · Accepted solution]

The cap remaining doesn't look at costop or contention in 5.x. If your cap rem is 5.5% and time rem is 1 yr, that is because your period of analysis is too short. The cap rem is last collection status, where as time rem is based on the period of analysis/trend.. e.g. bucket size and quantity.

If you want a better analysis. you'll need to attach a pdf of the view/report you're seeing so we can see all of the configuration policies.