Analysis of trends and variations
When generalising the FinOps approach, we do not only alert for the exceeding of certain thresholds, we also go further by analysing trends and variations.
For example, we can measure the costs per unit of measurement at regular intervals to check the trend by associating it with the perspective of the associated volume. In the example below we see that transactions are increasing while unit costs are stagnating, after a period of strong improvement (probably linked to our optimization efforts!).
It is also interesting to measure over time unused capacities and idle resources. In the example below we see an increase in July of Idle instances but the trend has been quite good since then. We also confirm here that the optimization efforts have drastically reduced the wasted compute even if we see that there is still room for improvement.