-
Notifications
You must be signed in to change notification settings - Fork 4
Description
Custom dataset in the DW for capacity reporting (tough one and time consuming)
[RB] what do you mean by custom dataset?
[DG] this is a long story :). When data is written into the data warehouse it either is added to a stanadrad dataset (alert, events, state, performance, alerts) or to a custom defined dataset (we have a few of them for Exchange, DPM and so forth). A dataset defines the data schema, aggregation and grooming model. Basically you need them when the data schema doesn’t fit the standard ones, or when you need a different aggregation model. Let’s say for example that we collect CPU usage for chargeback, we would need a sum of all the usage for a VM in a given period of time. Standard aggregations compute only avg, min, max, stddev while we need a sum for the period. Or let’s say we need some percentile for a measure, same story we need a custom dataset. This will take days to develop and tune and before that we should plan which measures we need and how we want them computed. Tough one.