Today Geoff and I met with Joseph Wolfgram, Russ[?] from UW School of Medicine and folks from Apptio. Apptio provides a service that takes your log data, asset data, ticket data (and more) and, to some extent, can discover your services but further, can show you how much your services/infrastructure cost in terms of capital, facilities, FTEs, and operations. It solves a very important problem.
The nice thing for UW Tech host/net management teams is that it fills a need that we generally don't currently fill and it sounds like it would integrate with our data/systems. We simply need to send our data to the service.
The meeting also showed me how others are looking at measuring IT costs/prices including FTE and operational costs. These ideas aren't particularly special but:
If operations support staff are ticketing all of their work, we should be able to measure their rate of ticket processing and, using the general cost of the positions, measure the cost per ticket. Thus, we could generally show how much an incident cost to resolve. Even using just the information we collect now, we could show within an order of magnitude what an incident cost.
Further, over time, we could include this number in our calculations of the cost of the service. E.g. over a quarter, we could see how much time was being spent on layer 2 incidents and put that in our calculations for the total cost of ownership for the service. It could go in to our per port calculations. Then we could perhaps calculate how much money we could save/lose if we chose more expensive switches with better management features.
Another opportunity could be to measure and then perhaps improve NOC employee ticket processing performance. If the average ticket processing rate per Tier 1 employee is X, then you can partially measure an employee's productivity by comparing their particular ticket processing rate to X. Certainly this measure would need to be taken with a grain of salt because without further instrumentation, it wouldn't take the quality of customer interactions into account and other important metrics but, still, it's still a useful measure. Also, it can provide a feedback loop for employees to see how they are doing compared to the average which, as they modify their behavior to stay average or better, would improve the average. This can all seem kind of Machiavellian but I think if it's used with care could really be beneficial for the quantity and quality of our operations service.