Getting a serious discount on cloud resources in exchange for a commitment lasting one or three years sounds great. You’ll need cloud capacity anyway, so why not buy it via GCP CUD (Committed Use Discounts)? If you’re negotiating a deal now, stop and read why there are better ways to reduce your cloud costs.
GCP CUD: let’s cover the basics
Google has had Committed Use Discounts on offer for the past few years. The system was designed to compete with AWS Reserved Instances without requiring advanced payments like with an AWS savings plan to get the maximum available discount.
You can choose from two types of GCP Committed Use Discounts: Resource-Based CUDs and Spend-Based CUDs.
Resource-based Committed Use Discounts deliver a discount if you commit to using a minimum level of Compute Engine resources in a specific region. This GCP CUD was designed for predictable and steady-state workloads. You can also use CUD sharing, which lets you share the discount across all projects tied to your billing account.
Example of a resource-based CUD: You get 100 vCPU for M3 in us-west1
- This applies to vCPU, Memory, GPU, Local SSDs, and images.
- You can use supported services such as Compute Engine, Dataproc, and GKE.
Spend-based Committed Use Discounts give you a discount if you commit to spending a minimum amount ($/hour) for a GCP product or service. This type of CUD was developed to help companies generate predictable spend measured in $/hr of equivalent on-demand spend.
Example of a spend-based CUD: You commit to spend $50/hour on Cloud SQL (Postgres) in us-west1
- This amount applies to aggregated cloud spend for cloud resources
- You get access to all the supported services like Cloud SQL, Cloud Run, VMware Engine, GKE Autopilot.
Many teams use CUD monitoring tools to make sense of their spend. For example, the CUD analysis dashboard helps to visualize and evaluate the effectiveness and financial impact of the GCP CUD you bought. The CUDs cost breakdown chart also comes in handy for monitoring CUD costs and understanding how much money you’re saving. Speaking of saving…
How much can you save with GCP CUD?
When you enter a committed use contract, you buy Compute Engine resources like vCPUs, memory, GPUs, or local SSDs at a discounted price. In exchange, you commit to paying for them for one or three years.
According to Google Cloud Platform, customers using CUDs can achieve up to 57% of discount for most resources like machine types or GPUs. With memory-optimized machine types, the discount can reach as high as 70%.
How do GCP Committed Use Discounts work?
Suppose you have a bunch of workloads that are stable and relatively predictable. You estimate that they’re going to stay the same for a while. This is the most common use case for GCP CUDs – all you need to do is commit to using a specific number of resources or spending a given sum of money over one or three years.
What do you need to commit to a level of service with Google Cloud Platform? You must have sufficient space in your quota. Quota limits the amount of a particular shared resource. For example, Rate Quotas restrict the number of requests you can make to an API or service, and Allocation Quotas limit the usage of resources without a rate of usage.
The idea behind quotas is preventing unforeseen usage spikes, and helping teams manage their resources better. You can always request Google to increase your quotas.
Once you commit to a contract, expect to be billed each month for the duration of your contract term. You’ll need to pay for these services, no matter if you use them now. Any services used above the commitment level will be billed using the higher on-demand rate.
What are GCP Sustained Use Discounts, and how do they differ from CUD?
Committed Use Discounts often get confused with Sustained Use Discounts. The latter are automated discounts you get on incremental usage after running Google Compute Engine resources for a significant part of a billing month.
The longer you run your resources continuously within the billing month, the bigger your potential discount on incremental usage.
Google calculates these discounts automatically and applies Sustained Use Discounts based on your usage. You’ll find them listed in your bill as a separate item.
4 risks of GCP Committed Use Discounts
GCP CUD might look interesting at first glance. After all, if you’re planning to use cloud resources, why not get the same resources you need for a lower price?
Think twice before signing a contract – GCP CUDs come with some serious drawbacks.
Changing requirements will cost you a lot
When you commit to given resources or levels of usage, you need to assume your requirements won’t change while the contract is running.
But a lot can change in the cloud world within a single year. Not to mention your business requirements! Even tech giants with entire departments dedicated to managing and forecasting cloud costs cannot make accurate forecasts.
Pinterest is a good example of that. Users spent so much time browsing gift and decoration ideas during one holiday season that the company’s cloud bill went way beyond the initial estimates. By that point, Pinterest had already committed to paying $170 million to AWS – and then needed to get extra capacity, as a result spending $20 million more than forecasted.
You risk getting locked in with the cloud provider
By entering into a contract with Google, you may be vulnerable to vendor lock-in. Over time, you might become dependent on GCP for its products and services within one or three years. What if Google decides to subset a service you’re using during your contract?
You may lose flexibility
As requirements change, so do your computing needs. When new challenges arise, you might have to commit to even more or get stuck with unused capacity that you’ve already paid for. Either way, you’re on the losing side.
You’ll have no flexibility of scaling or the ability to configure multi-region/zone distribution easily.
Setting up GCP CUD takes a lot of manual work
The DevOps approach requires advanced preparation, budgeting, and commitment to support seasonal usage fluctuations.
Selecting the optimal VMs that will meet your application’s demand and workload patterns today, tomorrow, and a year from now is complicated, to say the least. Manual methods are time-consuming and imprecise, leading to unreliable results.
Sure, you can use solutions for cloud bill analysis that can help you reserve the right instance type, but they might direct you to reserve capacity based on the instance type you’re using now rather than what’s right for your application. You might have issues like overprovisioning and cloud waste.
You don’t need to commit in exchange for a 60% discount
Instead of reserving capacity upfront for a discounted price, consider using cloud costs optimization techniques such as rightsizing, autoscaling, and spot instances. They allow you to stay flexible and reduce the cloud bill by ensuring that your teams provide just what they need.
We recently analyzed cost data from clusters under our management and discovered that teams reduce their monthly cloud costs by almost 50% by eliminating overprovisioning. By adding spot instances to the mix, companies cut their cloud bill by 60% on average.
If you’re looking for an alternative to GCP CUD, start with free Kubernetes cost monitoring and turn on automation that saves you 60% without any commitment.
CAST AI clients save an average of 63% on their Kubernetes bills
Connect your cluster and see your costs in 5 min, no credit card required.
Leave a reply