Organizations are being challenged to reduce capital and operating expenses—hardware purchases, software licenses, power consumption, and cloud costs—while keeping up with increasing workload demands. Cloud computing has played a significant role in how you can quickly scale up and down resources to meet business needs. However, the odds are good that your company has at least a few workloads living in a public cloud such as Google Cloud™ or AWS. Odds are also good that many of the instances those workloads use are the same generation as the day they were spun up – or several generations behind the latest. Significant cost reductions can result from improving operational efficiency through optimized performance and leveraging the latest technology.
Concerns about cost increases, a perception that it’s difficult to modernize workloads, confusion about which new instance to choose, and general inertia are all reasons why cloud workloads commonly remain on small, old, or free-tier instances. You might think that sticking with older, lower-priced instances is a no-brainer—why would anyone spend more than necessary? There’s more to it, however, and in this blog, I’ll explain how upgrading to newer instances can actually save you money.
Benefits of newer technologies
Before we dive into the ways that newer instances can save you money, let’s look at some other advantages they offer. As hardware manufacturers such as Intel® upgrade their products, they introduce improvements, including better security features. With breaches, the discovery of new vulnerabilities, and other security issues hitting the news on a regular basis, it’s clear that keeping your company’s workloads secure is vital. One way to do that is to ensure that you are using up-to-date cloud instances (hardware, software, drivers, and more) to take advantage of the latest security measures these updates introduce.
Other benefit are the new features that are introduced to help enhance performance. For example, newer versions of some Intel processors come with workload accelerators, such as Intel® QuickAssist Technology (Intel QAT) first released on 4th Gen Intel® Xeon® Scalable processors. This accelerator offloads some computationally intensive operations to increase core and thread efficiency.(1) To measure the performance impact of QAT, we ran web servers backed by NGINX on different Google Cloud™ instances, some based on processors with QAT and some without it. In our tests, an NGINX web server workload on the n4-standard-8 instance with 5th Gen Intel® Xeon® Scalable processors with QAT provided 1.41 times the performance and 1.44 times the performance per dollar of an n2-standard-8 instance with 3rd Gen Intel® Xeon® Scalable processors that do not have QAT.(2) Clearly, continuing to run your workloads on the older, lower-priced instances is a false economy. Many newer cloud instances also offer new and advanced features that older instances cannot use, such as Google Cloud™ Hyperdisk tier of storage(3) and higher network bandwidth limits.
Increased value
While new features and tighter security have obvious benefits, we know that decisions involving where to host workloads will always require balancing features, performance, and cost. Newer and better typically comes with a higher price tag, so the question becomes “Is it worth it?” Our tests of cloud instances say “Definitely.” Let’s say you have a natural language processing (NLP) AI workload running on an older AWS EC2 m5.8xlarge instance. It still functions adequately, but you’ve noticed that your cloud instance is running at its limit with no room for growth. You could spin up a second identical VM to share the load or upgrade to a larger instance in the same series, the m5.16xlarge; based on pricing as of November 2024, both of these options would double your monthly cost.(4) What about a third option, upgrading to the m7i.8xlarge? According to our tests and the AWS pricing list, you’d see 1.61x the performance(5) while paying only 5% more.(6) Put another way, you’d get just over 1.5x the performance per dollar.
It's always worth checking the prices of newer instances because their pricing is not always higher. For example, according to the Google Cloud™ hourly pricing list, newer C4 and N4 instances with 5th Gen Intel® Xeon® Scalable processors are actually slightly less expensive than their C3 and N2 counterparts, respectively.(7) This makes the decision to upgrade even easier. According to our tests, you could see a substantial increase in workload performance and performance per dollar when moving from N2 to N4 or C3 to C4. For example, on compute-intensive integer and floating-point workloads, the c4-standard-8 instance delivered up to 1.35x the performance of the c3-standard-8 instance and up to 1.38x the performance per dollar (see Figure 1).(8)
Figure 1: Relative estimated SPECrate®2017 performance and performance per dollar of C4 VMs with 5th Gen Intel® Xeon® Scalable processors and C3 VMs with 4th Gen Intel® Xeon® Scalable processors. Higher is better. Results may vary.
On MySQL workload testing, we saw that the n4-standard-8 with 5th Gen Intel® Xeon® Scalable processors achieved 1.25x the performance and 1.28x the performance per dollar of the n2-standard-8 instance with 3rd Gen Intel® Xeon® Scalable processors (see Figure 2). Compared to the n1-standard-8 instance with 1st Gen Intel® Xeon® Scalable processors, the n4-standard-8 showed 1.89x the performance (see Figure 3).(9) Since the instances are priced almost identically, you would see a similar performance-per-dollar improvement.(10)
Figure 2: Relative MySQL performance and performance per dollar that N4 VMs achieved compared to N2 VMs. Higher is better. Results may vary.
What this means for you
As you can see, even when a cloud instance has a higher price tag, you can get more for each cloud dollar you spend by investing in newer technologies. These increases in performance per dollar could mean many things for your company. You could consolidate two older instances into a single new one to reduce spending. Or you could upgrade to a newer instance family, but at a lower vCPU count to get similar performance at a lower cost. There are many advantages to revisiting older workloads and ensuring that you’re getting the best value.
If you’re wondering whether upgrading to new instances would be a hassle, rest assured that public cloud providers generally make the process very easy, with only minimal disruption to the workload. For AWS, you simply power off the instance, choose “Change instance type” in the Actions drop-down menu, choose the new instance type, and power the instance back on.(11) In some situations, changing your instance type can be more complicated, such as changing between Arm® and x86 processor types. Skipping generations in Google Cloud™—going from, say, N1 to N4—also requires some extra steps, but they provide in-depth, easy-to-follow documentation to guide you through the process.(12) You can also explore cloud options around machine images and disk snapshots that may provide shortcuts to help you quickly get up and running again on newer instances. If nothing else, this blog should convince you not to put any new workloads you bring to the cloud on “cheap” instances by default. Newer really is better.
Conclusion
Many are skeptical about the benefits of upgrading to new technologies and assume that an “if it isn’t broken, don’t fix it” approach is the most economical one. While that might be true in some situations, sticking with an older cloud solution almost always means paying more than necessary. To get the most for your cloud dollar, choose newer instances. Improve performance and performance per dollar while taking advantage of newer features and better security. Get started on AWS and Google Cloud™ instances of the latest 4th and 5th Gen Intel® Xeon® Scalable processor by signing in to https://aws.amazon.com or https://cloud.google.com/?hl=en today.
Google Cloud: How to update your instances in a managed instance group:
In Google Cloud, you can update instances in a managed instance group (MIG) or apply configuration updates during repairs by using the Google Cloud console, gcloud CLI, or REST:
Update VMs in a MIG:
- Go to the Instance groups page in the Google Cloud console
- Select the MIG to update
- Click Update VMs
- Select a new template to update to
- Select Automatic as the update type
- Click Update VMs to start the update
Update configuration updates during repairs:
- Go to the Instance groups page in the Google Cloud console
- Click the name of the MIG to configure
- Click Edit to modify the MIG
- In the VM instance lifecycle section, select Update the instance configuration
- Click Save
Apply updates to specific VMs:
- Use the update-instances command with the -instances flag
Apply updates to all VMs:
- Use the update-instances command with the -all-instances flag
You can also use the set-instance-template command to set up a new instance template for your group.
Choosing the lowest-priced VMs that meet the minimum performance can end up costing more than VMs running on the latest generation.
(1) https://www.intel.com/content/www/us/en/products/docs/accelerator-engines/what-is-intel-qat.html
(3) https://cloud.google.com/compute/docs/disks/hyperdisks#machine-type-support
(4) https://aws.amazon.com/ec2/pricing/on-demand/
(6) https://aws.amazon.com/ec2/pricing/on-demand/
(7) https://cloud.google.com/compute/vm-instance-pricing
(10) https://cloud.google.com/compute/vm-instance-pricing
(11) https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/change-instance-type-of-ebs-backed-instance.html
Notices and Disclaimers
Performance varies by use, configuration, and other factors. Learn more on the Performance Index site.
Performance results are based on testing as of dates shown in configurations and may not reflect all publicly available updates. See backup for configuration details. No product or component can be absolutely secure.
Your costs and results may vary.
Intel technologies may require enabled hardware, software, or service activation.
© Intel Corporation. Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its subsidiaries. Other names and brands may be claimed as the property of others.