In our previous post we looked at how to reduce our total cost of Google Cloud using best practices and opting for free Google Cloud services. That was only the tip of the iceberg. The cloud still has many hidden features that can be leveraged to optimise cost. In this post we will dive into some additional tips and tricks to do just that.
Following the analogy of our previous post, we have stored our data in a data lake and imported it in to our data warehouse. A next step in our process would be to use that data.
Optimise costs on Google Cloud through Computing
Commonly this task involves a wide variety of transformation steps and a lot of computing power. Big data systems come to mind with the necessity of many compute instances. Those compute instances can be quite expensive especially when opting for GPUs or TPUs. Luckily for us the cloud gives us the possibility to enable or disable the instances to our likings.
When we need additional computing power, we can add more GPUs and when load is decreasing, we can opt to remove compute instances. Maybe somewhat controversial, but in some cases it is actually cheaper to keep your instances running. Google provides sustained use discounts up to 30% on their compute instances. This means that the longer your instances run during a particular month, the larger the discount would be.
The following graph shows how your effective discount increases with use, up to 30%:
For more information about sustained discount, have a look at the sustained use discount page.
Let us assume that your organisation needs computing resources for a longer period of time to run business critical applications. These applications cannot inquire any downtime and you may be needing backup servers to make sure you comply to the necessary uptimes. As mentioned previously you will receive up to 30% discount each month for those resources if you keep them running, but you can do better. As the need for these resources will not change in the near future, you can choose for a committed use contract that could give you up to 70% discount.
A committed use discount (CUD) are spend-based or resource-based discounts, depending on the chosen Google Cloud service and therefore not only applicable to compute instances. When you purchase CUDs, you receive discounted prices in exchange for your commitment to use either a minimum level of resources or spend a minimum amount, for a specified term of one or three years.
What if you are absolutely not certain about the needed computing resources. What if an organisation only needs computing power when users upload a picture to their application. A workflow should then be started using computing resources to process that image for example. When having such a fluctuating demand of computing resources, organisations can go for Preemptible VMs obtaining a reduction up to 80% for that instance.
Preemptible VMs are computing resources that automatically shutdown after 24 hours. Additional shutdowns may occur during the 24h period the instance is running. Before shutting it down Google does send the instance a notification beforehand. This makes sure that the instance has enough time to save it’s state. Organisation like the one mentioned before could certainly live with these limitations.
Conclusion
We saw quite a few different possibilities to reduce our cloud spending and there are still many to uncover. You are now able to reduce the total cost of ownership of many projects. This will enable your organisation to obtain a higher budget for additional innovation use cases and a faster digital transformation.
Book a Personalized Google Cloud Demo
Want to see Google Cloud in action? Book a free live demo that is personalized to your requirements, challenges and needs.