02-22-2018 10:22 AM
Can anyone recommend a chargeback model that has been successful, where a central group charges back to the user community . We are looking to make it easy for users to adopt SAS. The usage would be for SAS on servers and PCs. The Software is purchased centrally and we want to charge back to the user community. If you have any documentation on your approach and pricing methodology this would be greatly appreciated. We are not looking to purchase any "chargeback" software but rather suggestions on how customers have been successful doing this internally.
Thanks for your help!
02-22-2018 10:34 AM - edited 02-22-2018 10:59 AM
Absolutely great question. Let me see if I understood correctly:
the most common practice is to charge-back based on resources usage, which idea is not very complicated: your budget on resources and maintenance can be split up between the tenants (e.e, departments) by percentages.
The application of this simple idea can be somewhat tricky, sometimes, but general idea is:
- If possible, you can create different application servers between your tenants, which already helps to reserve resources (disk or memory quotas, even CPU in some scenarios) for the tenants.
- If you cannot, at least your tenants to be charged should be united in defined groups.
- Then you can collect the usage of those resources. If you have quotas, they can use the limits or less of what within their boundaries, which will help to manage their expectations.
- The usage will be split, as said, in percentages and then you can, mathematically, define your charges.
Now, the question, is how to collect those resources: by monitoring your system.
And you can monitor at different levels:
On the last one, my favourite tool is the Enterprise Session Monitor (ESM) created by Boemska ( https://boemskats.com/esm/ ), which already includes in the latest versions a way to calculate those charge backs, for processes generated in your SAS Compute tier. Check out the demo video.
02-22-2018 12:04 PM - edited 02-22-2018 12:04 PM
In addition to JUAN S OCS's very good answer, you might consider cobbling something together based on actual SAS usage. Here is an ancient paper of mine that may still be serviceable if you go that route.
Measuring SAS Software Usage on Shared Servers with the RTRACE Facility.
Proceedings of the 29th Annual SAS Users Group International Conference.