Happy to help - at CRUK we have a number of model objects that are allocating costs to applications:
- Projects - Currently we have a temporary measure of a project to application mapping file until our PMO update the source document to add the application (or service) direct to the project. I maintain this mapping file on an ad-hoc basis when a new project has started and/or the application is created in our CMDB. The allocation is even on the model
- Infrastructure (Non-DB Servers) - There are a number of data sources which back the allocation from servers to application. The first being our CMDB. This can contain one-to-one, one-to-many or many-to-many mappings so we have an interim file from the CMDB which shows the server as the Parent and the application as the Child then do a lookupex from the server data set to this relationship file. We have created a server allocation for the weighting which is Memory x CPU
- Infrastructure (DB Servers) - We use a secondary file which shows the DB server mappings to application from our DBAs with the DB Server to application mappings, again we use this file as a lookup to pull the data through to the Servers Master Data. The difference from the above is that the server allocation for these is the amount of GB size the application is taking up on that particular DB
- Tickets - This one is pretty straight forward as we get the tickets from our CMDB and then use the affected CI.
- Cloud Compute - Currently we're pulling this out at the tower level and mapping the costs directly on the model to our cloud applications. We are working with Apptio to get Cloud analysis in to our project where we can then remove this and have them being mapped to the Cloud instances directly at the Servers object then on to their rightful applications (if tagged in AWS!)
- Towers (People) - We have two allocation lines from towers to applications. 1 where we have pre-determined what the application is. We have done this a couple of ways, firstly we have mapped everyone who works on an application to either one or many applications and then attach the application to the key to allow that headcounts time to be sent direct to an application (if not allocated to Compute). Secondly, we use a file called the Order Tracker which our purchasing team captures all the orders raised and these are directly mapped to towers and applications - again making sure the orders are sent direct to an application (if not allocated to Compute)
- Towers (Non-labor) - The remaining allocations to the Application Tower where a specific application has not been defined we are mapping directly to an application called Unknown App. This is because we believe everything that is mapped to the Application tower should have a defined application. This Unknown App is then looked at monthly to see if anything has changed or can be updated on the order tracker, labor files or even in the CMDB (i.e. order tracker has identified a new application which isn't in the CMDB).
Hope this helps - apologies for the long post but I started work way too early this morning and have had an excessive amount of coffee