- Why Change?
Discover the power of cloud cost intelligence.
Give engineering a cloud cost coach.
Learn more about CloudZero's pricing.
Request a demo to see CloudZero in action.
Learn more about CloudZero and who we are.
Got questions? We have answers.
Speak with our Cloud Cost Analysts and get the answers you need.Get in touch
How SeatGeek Decoded Its AWS Bill and Measured Cost Per CustomerRead customer story
Enable engineering to make cost-aware development decisions.
Give finance the context they need to make informed decisions.
Decentralize cloud cost and mature your FinOps program.
Discover the best cloud cost optimization content in the industry.
Browse helpful webinars, ebooks, and other useful resources.
Learn how we’ve helped happy customers like SeatGeek, Drift, Remitly, and more.
5 Tactical Ways To Align Engineering And Finance On Cloud SpendRead blog post
This guide covers what On-Demand Instances are, their application, pros and cons, and when to use these types of instances.
According to Amazon, On-Demand Instances are best reserved for short-term processes with uneven workloads that cannot be interrupted. This might partly explain why such EC2 instances are exceedingly popular with AWS users.
On the flip side, however, it turns out that On-Demand Instances are also getting dumped for the sake of cloud cost optimization. Our recent Spot Instances guide, for instance, advises AWS users to consider making a switch and, consequently, enjoy up to 90% cost savings. Otherwise, you could alternatively opt for Reserved Instances, which have also proven to offer discounted rates.
But then again, don’t be quick to write off On-Demand Instances. While cloud cost optimization heavily favors Spot Instances and Reserved Instances, your organization may still need On-Demand Instances on some of its operations.
This guide provides more clarity by explaining what On-Demand Instances are, their application, plus the resultant pros and cons that you should expect. What’s more, you get to compare On-Demand Instances vs. Spot Instances, as well as On-Demand Instances vs. Reserved Instances.
Table Of Contents
On the AWS, On-Demand Instances happen to be the primary EC2 deployment model. This is the pricing option that allows you to instantly purchase uninterrupted Amazon cloud computing capacities using pay-as-you-go.
The pricing itself is based on an hourly or per-second basis, and you’re only billed for the EC2 instances that you actually utilize.
It’s worth noting, however, that while the prices are fixed, the precise charges per instance are not uniform — rather, Amazon’s On-Demand pricing schedule spells out different rates according to server region and EC2 instance type.
You’ll find, for instance, that T3.large instances have varying On-Demand pricing rates compared to A1.large or T3.nano. The same applies to US-based On-Demand Instances versus deployments in Asia and Europe.
Although the numerous variations might feel overwhelming at first, the good thing about On-Demand Instances is that they don’t require a lot of preliminary analysis. The computing capacity that you choose at first is not fixed — you can freely upscale and downscale it by adjusting the instances based on your application needs.
What’s more, Amazon’s On-Demand Instances offer uninterrupted control over their lifecycle. That means you get to choose when to deploy, terminate, reboot, start, or hibernate the EC2 instances — without being compelled to commit for the long haul.
In simple terms, On-Demand Instances are Amazon’s way of giving you the chance to freely purchase or scale any amount of EC2 capacity based on your organization’s real-time needs.
However, all this flexibility comes at a cost. On-Demand Instances happen to attract the highest rates among Amazon’s EC2 pricing modes.
While there are many similarities between On-Demand Instances and other popular types of AWS EC2 instances, it’s the distinct differences between them that determine when and how to leverage each category.
There are specific occasions, for example, where Spot Instances tend to be more appropriate than On-Demand Instances and vice versa. Then when it comes to long-term commitments, you might be better off with Reserved Instances.
Here’s how they compare:
Unlike On-Demand Instances, Spot Instances are not readily available for purchase. Instead, Amazon offers them from time to time as spare EC2 instances.
Users are essentially required to place price bids as a way of scrambling for any available Spot Instances when the opportunity arises. But, there are no guarantees — you only get to access the spare instances when their market price drops below your bidding rates.
This market price is technically known as the Spot Price. And, as it turns out, the rates are always shifting based on real-time market demand and the long-term supply of Spot Instances. More unused EC2 instances translate to lower Spot Prices, while a drop in the volume of unutilized EC2s affects the market prices negatively.
One consistent trend you’ll notice here is, Spot Instances are always cheaper than On-Demand Instances. Regardless of the direction the market takes, the spare EC2 instances come with huge discounts — potentially stretching up to 90% of what you’d have otherwise spent on On-Demand Instances.
That makes Spot Instances a worthwhile consideration when it comes to cloud cost optimization. But, don’t expect the discounted EC2 instances to take over all the functions reserved for On-Demand Instances. Spot Instances can’t match up in terms of flexibility or reliability.
You’ll notice, for instance, that Spot Instances are prone to interruptions. Amazon happens to have all the rights to terminate and take over your EC2s instances once their market price rises above your bidding rates. Even worse, the accompanying pre-termination notices only give the affected users a lead time of two minutes.
Combined, these attributes mean that Spot Instances would only be an ideal cost optimization option when you’re dealing with:
That said, the best approach here would be leveraging On-Demand Instances and Spot Instances concurrently. Such a complementary architecture would maintain the minimum required level of primary computing resources while, occasionally, taking advantage of the extra power that comes with cheap Spot Instances.
There’s no physical difference between Reserved Instances and On-Demand Instances. Both provide access to pretty much the same EC2 computing capacities.
But, not under similar pricing rates. If you compare their cost schedules, you’d quickly establish that Reserved Instances are cheaper than their On-Demand counterparts. And while the discounts might not be as huge as the Spot Instance offerings, they’re still quite significant. Amazon itself states that the pricing can dip by as much as 72%.
This cost difference is brought about by variations in commitment between the two categories. While On-Demand Instances can be acquired and dropped at any time, Reserved Instances will keep you committed for the long haul.
“Reserved” here means that you get to keep and run your EC2 capacity for a fixed duration, within which you’ll enjoy all the bells and whistles that typically come with on-demand computing. Then, in return, Amazon reduces the usage charges well below what you’d have paid with On-Demand pricing.
Now, the reservations here run for one or three years. Then since the discount rates are directly proportional to the commitment period, three-year Reserved Instances tend to offer the best cost benefits.
Unfortunately, such long commitments are also known to make the reservation process exceedingly challenging. It takes a lot of calculation and analysis to accurately predict the amount and type of computing resources that your organization will need over a three-year period.
As such, you could say that Reserved Instances would not be ideal for dynamic applications with inconsistent needs. Even when you happen to settle for the slightly flexible Convertible Reserved Instances, your best bet would be with:
By now, we can agree that On-Demand Instances offer some sort of a double-edged sword. While they have a number of solid benefits, On-Demand Pricing also comes with its fair share of weaknesses.
Based on the attributes we’ve established about On-Demand Instances, we can conclude that they are exceptionally suitable for:
Now that we’ve explored the basics of On-Demand Instances, you might want to revisit our previous insights on Spot Instances and Reserved Instances for a more in-depth comparison of the three. With that, you’ll gain more clarity on the Amazon EC2 pricing model that suits not only your business structure and workload, but also your cloud computing budget.
Don’t get fixated on one, though. The goal here is not settling for either On-Demand Instances, Spot Instances, or Reserved Instances. Rather, you should try to optimize your cloud costs by seamlessly blending all three of them. Strategic cost management entails maintaining just the right volume of On-Demand Instances while, at the same time, running less critical workloads on Spot Instances and Reserved Instances.
The downside to this strategy is, it requires a lot of ROI calculations, complex analysis, and consistent cost tracking. Thankfully, using a solution like CloudZero, you can understand where your cloud spend is going and make informed cost optimization decisions.
With CloudZero, organizations can drill into cost data from a high-level down to the individual components that drive their cloud spend — and see exactly what AWS services cost them the most and why. to learn more.