Calculating the Profitability of Azure Reservation for Workloads: Does it Always Pay Off?
Here at 2bcloud, we love a Cloud budget optimization query. We recently helped a customer in optimizing their costs for a cloud migration and wanted to share the conclusions and how-to with our readers.
Our use case was Reservation for one of our client virtual desktops for company employees. In this case, the customer was working with Citrix.
The requirement is fairly straightforward, they use 100 virtual desktops for 9 hours a day, multiplied by 21 working days a month.
After crunching the numbers, we discovered that the cheaper option, in this case, would actually be Pay-As-You-Go (PAYG) over Reservation prices!
Here’s how we got our results for the customer:
To calculate your PAYG cost, you basically need to get the monthly usage hours and multiply them with your hourly instance price. In our case:
9 hours a day * 21 working days = 192 Hours
192 multiplied by 0.097 (PAYG instance price per hour) = 18.624
We then compared the same but with the discounted 1-year reservation hourly price, and multiplied by a full month of usage hours as that’s how Reservations are accounted for:
720 * 0.0574 (1-year reservation instance price per hour) = 41.328
In order for Reservation to reduce costs, the PAYG price needs to be higher.
After running the numbers, we found that the point it becomes profitable to choose Reservation over PAYG for such use case of partial workload hours is at around 430 monthly hours. This may vary between instance types because of factors including discounts.
However, as discounts usually not vary too much, this calculation can be used as a ballpark value to estimate for future use cases.
If you’re looking for advice on optimal cloud profitability, schedule a call with one of our experts.