The Stack Archive Interview

Q&A: Why Reserved Instance optimisation is the future of cloud cost management

Mon 15 Oct 2018 |

Frank Wang, Product Marketing Manager for Densify, explains the challenges of cloud cost management, how Reserved Instances (RI) can help cut costs, and what businesses should consider before committing to an RI offering.

Why is cloud cost management becoming such a headache for enterprise customers?

Cloud operations teams are stuck between a finance department seeking to drive down cloud costs and application owners demanding more cloud resources. On top of that, matching workload demands with the right public cloud products is complex: demand patterns are hard to identify and there are simply too many cloud service combinations to choose from.

How can the use of Reserved Instances help to alleviate this challenge?

Reserved Instances (RI) are like coupons that help you reserve and save cloud costs, but if mistakenly used can waste your money in the long run. Although they offer discounts, because they lock businesses into a specific instance type and size for one to three years, they reduce flexibility.

RIs are a commitment, so it is very important to get it right and commit to the right thing. Reserving the wrong instance not only costs businesses more money: it also introduces operations risks and performance issues for cloud applications.

What should businesses consider before committing to an RI offering?

Businesses should always reserve based on their optimised set of RIs, not their current configuration. Most enterprises are aware of the benefits of RIs, and many vendors will recommend businesses RI plans to help them save costs. However, their purchasing recommendations are based on the current set of instances, not optimised ones.

Suppose an organisation is currently using 100 m4.large EC2 instances from Amazon and a vendor makes an RI recommendation based on this configuration. Clearly, this is nonsense. What if those 100 m4.large should be a combination of c5.large, m4.large, t3.medium, and i3.large instances? The vendor has just locked the organisation into a costly RI and put its cloud performance and operations at risk. Densify uniquely employs a multi-pass strategy: we optimise first, before committing to a reserved instance.

Businesses should also ensure they have an optimised RI purchasing strategy. Not every instance in use should be reserved. If the predicted uptime is not high enough, it is better off to just use it with an on-demand option. Why waste money reserving the things you will not be frequently using? This optimisation should be built-in.

Additionally, if a business’s existing set of RIs need to be optimised, they need to ensure this transition is smooth and intelligent. We take existing RIs and analyses expiry dates, convertibility and other factors, to come up with a month-by-month rollover plan to take RIs into the optimised state.

What is the role of machine learning in all of this?

Driving all of our optimisation is machine learning, it is what separates ordinary RI optimisation from intelligent RI optimisation. ML algorithms can identify the perfect match and find out the optimal instance type and size, they can analyse and recommend the optimal purchasing strategy of RIs, and they can be used to analyse the expire date, convertibility and other factors to offer lifecycle management and take a fleet of RIs to the optimal state.

As its value becomes clear, ML will become foundational for informing RI purchasing decisions. If a business’s RI choices are not informed by machine learning they will soon be trailing the rest of the pack.

To learn more around optimising your Reserved Instances, download Densify’s definitive guide below:

Please fill out your details below to download this whitepaper:



First Name:

Last Name:

Job Title:


Select an Industry:


By ticking this box, I consent to the sharing of my details with the related third party whose products and services may be of interest to me.

Error: Contact form not found.


Cloud cloud management densify White Paper
Send us a correction about this article Send us a news tip