How To Save 50% On Your Cloud Bill With CAST AI

This article is sponsored by CAST AI

A typical cloud bill goes 23% over budget (Flexera data). And no wonder it does when you have to control all these things:

  • Dozens of resource types
  • Resources you might not need
  • Shadow IT projects
  • And many more

Cloud providers aren’t helping you unpack all these costs either with bills reaching hundreds of lines.

This lack of clear visibility into your costs will prevent you from optimizing your infrastructure so that it supports your teams but also keeps costs in check.

Analyzing a cloud bill manually is difficult and time-consuming because each service has its own billing metric and cost insights are spread around the CSP console. This is where automation can help.

Here are three ways CAST AI helps teams to save on their cloud costs and get best-of-class features across multiple cloud services.

1. Choose the right VM instance

Compute resources are the most expensive item on your bill. By selecting the right instance type, you stand to save even 50% on your costs.

But doing that manually is hard. There are so many options available even within a single cloud service provider.

This is a perfect use case for automation. The CAST AI optimization algorithm is always on the lookout for opportunities to replace your VMs with less expensive alternatives that do the job, also in a multi-cloud setup.

How does this work in practice?

We tested this cost-savings approach on an open-source e-commerce app. We load-tested the application with a high number of concurrent users and generated a likely 30-day usage pattern.

Originally, we selected the m5.xlarge VM that comes with 4 CPUs, 16 GiB of memory, and up to 10 Gbps of network bandwidth. The cost of this instance type on-demand is $0.192/hour.

We launched CAST AI and, using automated analysis, the optimizer selected an alternative shape: a1.xlarge, an instance type with 4 CPUs and only 8 GB of RAM. The application could easily fit into it which means we wasted RAM in our first choice. The new instance had a reduced network throughput, but this wasn’t an issue for the app.

The best part? The a1.xlarge instance costs only $0.102/hour. This means that we instantly saved 46% per compute-hour.

2. Take advantage of autoscaling

Another tactic for lowering your cloud bill is rightsizing your resources via autoscaling features. CAST AI optimizes your costs because it dynamically aligns the resources to your workload demands.

How did that play out in the case study?

This chart shows a part of the distribution of usage of the app over 30 days (720 hourly data points, with data aggregated to the hour). To cover the usage with its spikes, we’d need to use 5 VM instances at all times. But 5 VM instances are only required for a small amount of time within that 30-day window. This is where autoscaling helps.

Without autoscaling, the optimized VM instance brought the total compute cost to $367.20 over 30 days. By using autoscaling, we reduced compute costs to reach $197.37 over 30 days – an additional savings of ~45%.

3. Use spot instances instead of reserved capacity

Reserving cloud capacity is a path to vendor lock-in. But there’s another option to lower your costs: spot instances.

Spot instances are the unused capacity providers offer at much lower prices, even up to 90%. The catch? They can pull the plug whenever someone else orders these resources and they give you short notice, from 2 minutes to as little as 30 seconds. Spot instances are more difficult to manage for production workloads – unless you have automation in place.

The final step in our case study

The initial monthly cost of running the app on the unoptimized VM instances was $691.20.

We applied the CAST AI spot instance policy to save on costs relative to the on-demand pricing.

When using the most aggressive policy settings – where all the instances in use are spot instances – we brought the compute costs down to $65.01. We saved 90% over the original costs of the unoptimized deployment.

Find out your cloud savings potential.

With savings like above and the technology to get them within reach, there’s no reason to keep waiting.

Discover your potential savings using the Savings Calculator and pass the results forward to your DevOps team!

This is a special advertising section. The content herein contains the views and opinions expressed by the advertiser.