Unsolved
This post is more than 5 years old
3 Posts
0
144049
March 26th, 2013 20:00
Foglight Chargeback
All,
I am playing with Chargeback. I have a mandate to complete with our parent company on determining how much we spend in terms of money, electicity and carbon footprint per eCom transaction.I am hoping Chargeback is a platform to build on.
Just throwing some questions out there. Feed back and general input is certainly welcome -
1) Has anyone done anything like this before? Can they provide some specific insights?
2) To what degree can chargeback be automated out of the box. For example, I don't want to manually create any groupings. When new ESX and physical servers are created I expect that they are auto-added and auto-removed.
3) How does Chargeback handle removed objects? For example, I have a box tx-apache01 whichi s 8 CPU, 16gigs of ram. It dies, I replace it with an object with the same name which now has 16CPus and 32gigs of ram and I delete the old foglight object.
Any other areas you think I might be of interest?
DELL-Thomas B
171 Posts
1
March 26th, 2013 21:00
Normally what I see is most people look to do chargeback based on an allocation method for CPU/Memory/Disk and then a column for environmentals (HVAC/Power/Rack&Stack/etc...) and FTE costs. Beyond that you can get fancy at allocation costs for software, better monitoring/backup support etc... but those are rarely cost accounted for. Those do require an understanding of what the total costs already are and then dividing them based on where we 'think' the costs should come from. X dollars for Memory, Y for Storage and Z for CPU etc... and then breaking out with the ideal ration of VMs. "We thing we can fit 10:1 per host so the breakout should be X/10 etc.."
If you wanted to compute based on actual power consumption live (the metrics are in vCenter via plugins or via hardware) it could be done but it is quite a bit more work (custom) to get that to happen. Doable, but probably a few days.
As for the systems, most group objects by tier somehow. All VMs in Cluster A are Tier1/Gold or All Datastores called Prod are Tier 1 for storage etc... Then you tie the tiers to cost and the groups to the tiers and away you go.
dpw1
3 Posts
0
April 22nd, 2013 21:00
Thanks for the feedback. Ending up using another monitoring solution for this.