Pricing of services on AWS cloud comes in different flavors such as free tier, pay as you go, spot and finally commitment based pricing models such as reservations and savings plan.
The commitment based pricing models can lead to considerable savings, if utilized optimally for the right use case. However, it is important to understand the practical considerations when opting for this pricing model.
This article summarizes few of the key tips to keep in mind if planning to opt for this pricing model. Please refer to my whitepaper for more details on each of the below items.
Over-assuming savings percentages
The highest savings percentages often comes with riders such as longer commitment periods, stringent usage terms such as being restricted to only a particular service type and higher upfront payments. Be pragmatic with your organization's risk appetite and savings that can be obtained based on that constraint.
Crossover points vs breakeven point
You may be asked to present a forecasted pricing plan and savings when proposing this pricing compared to a pay as you go model. It is important to understand when the crossover happens and when the actual breakeven occurs and not confuse the two.
Crossover point is when the pricing of the commitment based model matches the pay as you pricing and subsequently gets cheaper. The catch here though is you have still committed to a set usage period in case of commitment based pricing whereas in pay as you go pricing, you are under no such restriction to continue using. An actual breakeven point calculation will need to take this into perspective.
Explore alternative commitment options
For shorter term commitments, there are sometimes better alternatives than directly opting for a reservation or savings plan model, both of which come with a minimum of 1 year commitment such as exploring the reserved instance marketplace. An example is purchasing off the reserved instance marketplace where there is an instance available with lesser commitment period remaining. Explore these alternatives beforehand.
Thoroughly understand your workload patterns
The Savings plan option is usually the more flexible of the two options but that does not mean it is also always the right option. It may turn out expensive for certain patterns and also if you need to cancel or dispose of the commitments at short notice. For example, With reserved instances you have option to sell them off on the marketplace but not with savings plan. Be sure to weigh the pros and cons of both options for your workloads.
Understand price calculation difference
The Savings plan bills you on hourly commitments whereas Reservations are billed on an a monthly basis. Whereas this may not seem to be big deal at first glance, it can very well turn out to be one when there are large fluctuations in workload usage. A key point to understand here is that with savings plans, if your workloads consistently under utilize the committed savings plan amount, the loss in commitment amount cannot be adjusted in retrospect in subsequent period maybe for some spike usage in future.
So analyze your workload usage patterns before opting for either options.
Miscellaneous considerations
Other than the above tips, there are also items to look at. Some of these include,
Pricing changes of the underlying services
Cloud providers such as AWS regularly come up with price reductions for their services. For example an EC2 instance type that cost x USD/hour sometime back might now cost .95x USD/hour in pay per use mode. So if you have calculated the savings of commitment based pricing over pay as you go assuming x USD/hour as the pricing for pay as you go, that calculation needs to be reworked.Calculate savings based on practical usage time
When comparing commitment based pricing with pay as you go model, understand that with pay as you go model, you are under no restriction to use the service, 24x7. But with most commitment based pricing, you sign up for a fixed hourly or monthly commitment price, irrespective of usage.
So account for this when comparing these pricing models. Also, below a certain threshold of usage, it may not make financial sense to commit to a commitment based pricing so understand your practical workload usage patterns.Introduction of newer instances at better price points
Cloud providers regularly introduce a new class/series of instances, at better performance but lower price point than the existing instance class. Based on how flexible the commitment option is, you may be able to take advantage of this by switching to the newer class of instance.Centralizing your reservation strategy
For multiple teams and accounts, centralizing the process will lead to better usage visibility across teams and greater organizational efficiency.
To conclude, it is important to assess and plan out your commitment based pricing journey carefully, not over-commit initially and adaptatively optimize the commitments made based on past and forecasted usage data.
Top comments (0)