Into the Deep End: AWS Pricing and Cost Savings Strategies

Who hasn't jumped into the deep end of the cloud by now? Something like 94% of organizations use cloud in one form or another and, every year, more IT budget goes to cloud. That progression makes it imperative for you to understand the ins and outs of cloud spending with any cloud provider you use - Amazon, Google, Microsoft - even Oracle or IBM. You're in a spot where you need to get the most value out of every dollar, and that's not easy.

Let's consider AWS for a minute. AWS pricing optimization is a discipline unto itself with many proven practices, frameworks, models, and even dedicated tools (like NetApp^® Cloud Insights). AWS, like other clouds, has many levers to pull and knobs to turn that increase and decrease costs. It has many choices you can make that have ongoing implications on spending, and many ways that simple misconfigurations or misuse can add up to cost overruns over time.

At NetApp, we're here to help, so let's solve some of those AWS cost challenges. In this post, we're going to give you an understanding of AWS pricing models that can reduce your costs. We'll also provide 10 reasons why your cloud bill might be different than expected. Finally, we'll show you a powerful way to reduce your Amazon Simple Storage Service (Amazon S3) costs by using NetApp Cloud Volumes ONTAP^®. By the time you've finished reading this post, you'll have new ideas and new resources to drive down your AWS costs - leaving new room in your budget to adapt and innovate.

Ready to jump into the deep end?

AWS pricing: models to make your business better

On-demand. We're familiar with the on-demand pricing model - it's what we think about when we first think of cloud. On-demand is the default AWS pricing model, based on actual utilization and billed by the hour or the second. It's easy and flexible, but it's almost always the most expensive approach.

On-demand pricing is a good fit if your organization prefers to avoid upfront costs or time commitment. It's also a good idea if your applications are mission critical or have unpredictable load spikes.

Most organizations begin with on-demand pricing and then shift to another model when they have a deeper understanding of their goals and requirements and are under new pressure to cut costs. There's nothing wrong with starting with on-demand pricing, and it's the right fit for some workloads. But continuing with it for every requirement will result in higher costs than some alternative models.

On-demand enterprise discount program (EDP). If you're committed to long-term on-demand utilization, you should apply for an Amazon EDP. In exchange for your purchase of enterprise support and a commitment to a set spending for a period, an EDP gives you a substantial discount. It's also possible to negotiate deeper discounts with an EDP, especially if you do some footwork - mapping out your future workloads and consolidating everything your organization does into a master account.

An EDP works well if your organization has a large cloud footprint and a large budget. It also works well if your applications aren't rapidly growing or spiking unpredictably. With an EDP, we've seen discounts of up to 13% on a $5 million monthly expenditure.

Reserved Instances. If you are willing to reserve instances for 1 to 3 years, you can enjoy significant cost savings with Reserved Instances (RIs). But there is a trade-off - RIs lock you into a long-term (1-to-3-year) fixed-cost model with a set number of instances. Even if you reduce your workloads by 50%, you cannot reduce your instances. Conversely, if you need instances beyond your reservation, you'll have to choose to use on-demand resources at a higher cost or choose more Reserved Instances.

RIs are a good fit for organizations running predictable existing or legacy applications in the cloud. Reserved Instances offer discounts of up to 75%.

Spot Instances. Only available for Amazon Elastic Compute Cloud (Amazon EC2), Spot Instances offer the best discounts but are available only through a complex mechanism. You use Spot Instances by participating in an auction of unused compute capacity on Amazon's open market. Pricing can change every few minutes and, if you place a bid that's above the current market price, you receive a Spot Instance. However, Spot Instances can be interrupted by the Amazon Spot service if capacity isn't available or if the marketplace price goes higher than your bid.

As you can probably guess, Spot Instances aren't a good fit for traditional workloads or many organizations. But if you have the expertise and tools needed to use stateless, highly distributed applications, Spot Instances might work for you. To effectively use Spot Instances, you need to have expert cloud-native developers who can take advantage of the benefits while dodging the disadvantages. Spot Instances offer up to 90% discounts when compared with on-demand EC2 pricing.

Let's jump into the deep end of a different pool for a few minutes. Now that you understand four basic AWS pricing models, we'll explore another topic: reasons your AWS pricing might fluctuate.

Avoiding costly surprises: 10 reasons your prices might vary

If you're familiar with the AWS pricing calculator, you know that it paints a picture of costs that are always under control. In reality, most organizations see significant variation in their month-by-month cloud costs. It's normal, but certainly frustrating. Are you ready to learn why?

1. Actual usage isn't the same as predicted usage. Without firm limits, your organization might be using more instances or resources than predicted.
2. Your teams might also use different services than predicted. As development progresses, it's common for teams to discover new services that make their projects better.
3. Predicting data transfer and throughput costs is an art that few organizations have mastered. These costs are hard to predict, difficult to track, and easy to overlook because only some Amazon services charge for data transfer or throughput.
4. Instances in one region cost more or less than another region. If you have multiregional operations or are using more than one region for disaster recovery, prices will vary.
5. You had a good lock on prices in June - but in July, everything's different. Amazon changes pricing, so you're likely to experience some level of unpredictability. These pricing changes, of course, become even less predictable as you use more on-demand services.
6. Amazon often doesn't include taxes or other government charges in its calculators - don't be surprised when you receive your invoice.
7. If you've recently moved to Amazon, you might have benefitted from its free tier or new customer discounts, which subsequently went away - increasing your prices.
8. Sometimes Amazon calculators do not understand tiered offerings, especially if you move data from Amazon S3 to Amazon S3 Glacier. NetApp offers a storage-focused AWS calculator to help you determine the impact of storage tiers.
9. Third-party licensing fees for commercial software, whether installed by you or purchased through the Marketplace, might not be included in your calculator estimations.
10. Finally, the calculator does not take into account daily fluctuations for exchange rates. Your pricing will fluctuate if you're paying in a different currency.

Clear as mud?

A tool that can help: reducing your overall cloud costs

OK, we've covered pricing models and cost fluctuations, so you're armed with many new ways to address your AWS costs. Here, as we're wrapping up, we're going to bring another cost-control measure to your attention.

NetApp recognized that cloud storage costs were a major contributor to cost overruns, and our customers asked for help. That's why we created NetApp Cloud Volumes ONTAP. It's an enterprise-grade, secured, proven storage solution for many clouds - including AWS.

With ONTAP, you can unify and simplify up to 368TB of cloud storage for a wide range of use cases: file services, databases, DevOps, or any other enterprise workload. It offers all the features you'd expect from NetApp (high availability, data protection, storage efficiencies, cloud automation, Kubernetes integration) without impeding or affecting your access to other cloud services. ONTAP can be used with a single cloud or serve as a bridge for a multicloud infrastructure. Thanks to efficiency features such as thin provisioning, data compression, and deduplication, you can reduce storage costs by up to 70%.

We hope all this information will help you move forward with better cost controls and less uncertainty. To learn more about NetApp Cloud Volumes ONTAP and Cloud Volumes Service, visit NetApp Cloud Central.