Over the last few years, an increasing number of companies have been moving from on-premises to the cloud. In the early stages of cloud adoption, users tend to focus on the product development while operational efficiency usually takes a back seat. However, once their product or services have stabilized, users will start paying attention to cost concerns and begin looking for possible ways to reduce their cloud costs.
With just a click of a button, you can start using the cloud and will be charged only for what you use (pay-as-you-go model). However, it is so easy to increase the resources causing an upsurge in the usage fee. It is an increasing concern for many users that cloud usage and its cost have become a black box.
Our customers often ask us for better ways to reduce cloud costs without sacrificing performance. In this article, we will share our recommendations on computing resources and GCP cost reduction measures for IaaS.
1. Identify GCP projects that are costly
- First, understand the overall cost and see which GCP projects and services are contributing the most to the overall cloud spend.
- GCP provides a free Cost Management tool, from which you can see daily trends and costs over any period of time.
- Let’s decide what to do first with the project starting from the one(s) with the highest cost.
- 2. Detect idle computing resources and reduce them
- a. Check the status of running Virtual Machines (VMs)
- Check for idle VMs and stop them to help reduce costs. It can be set it to start and stop automatically if the VM is not in use on a certain day of the week or in a certain period of time, e.g. on Saturdays and Sundays, or in the night time.
Recommender in GCP allows users to see the idle VMs and subsequently optimize the resources.
- b. Use a rightsized VM
- VMs are often selected with sufficient CPU and memory so that they don’t run into system problems. However, in case of capacity underutilization, it will most likely result in wastage.
You may reduce the costs by changing VMs to the ones with appropriate size according to machine type recommendations.
- c. Leverage preemptible VMs and Spot VMs
- GCP has affordable VMs called preemptible VM instances and Spot VMs, that can be used at a low cost while maintaining the same level of CPU and RAM performance.
Spot VMs are the latest version of preemptible VMs announced in October 2021 and are currently in Preview stage, and it seems that it will be generally available in 2022.
Discount rates and maximum instance uptime differ as follows.
|Preemptible VM Instances||Spot VMs|
|Maximum Instance Uptime||24 hours||No Limit|
- * The preemptible VMs’ discount rate will become the same as the Spot VMs in the future.
Currently Spot VMs are available in limited regions and machine types.
- There are also some points for you to keep in mind when using preemptive VMs and Spot VMs.
- i. Preemptible VMs and Spot VMs can be shut down at any time
ii. It is not always possible to secure a VM because it uses a surplus VM on the market
iii. The price (discount rate) is not fixed by the type of VM, instead changes from time to time.
iv. Not covered by Service Level Agreement
Keeping in mind that it may be shut down at any time, if utilized well in the development or testing environments, it will lead to a significant cost reduction.
- d. Take advantage of committed use discounts
- If you cannot use preemptible VMs and Spot VMs because it can be shut down anytime, you can get a discount by using committed use discounts.
- You can get a 25% discount for 1 year or a 50% discount for 3 years by specifying the region, machine type family, etc. of Google Compute Engine and committing to use it for 1 or 3 years.
- Upon purchasing, whether or not the resources are used, you will be billed monthly for the resources purchased for the duration of the term selected. If you have a predictable workload, you can cut costs on resources that you need by using the committed use discounts.
- Committed use discounts recommendations are also displayed in Recommender to help you understand what you commit and how much you can reduce.
Let’s check the following points for GCP cost reduction that can be implemented from today.
This article describes the cost reduction of computing resources, which usually account for the majority of cloud costs. Cost optimization can be achieved by making good use of the purchase options provided by each cloud vendor.
Alphaus provides tools for managing, analyzing and optimizing public cloud costs. Please feel free to contact us if you are experiencing challenges in cloud management or want to reduce cloud costs without sacrificing performance.
Principal Customer Success Manager