A Guide to Understanding Your GCP Account Costs
Christina Harker, PhD
Marketing
Google Cloud Platform (GCP) has a reputation of having an easy-to-understand pricing model. However, if you’re new to GCP or don’t have a background in Cloud technology, it can still be a little overwhelming. GCP has a wide range of services and functionality, each affecting the price you pay.
The good news is that GCP’s prices can be broken down, ready to be analysed and have budgets set. In this guide, we’ll go through what GCP costs are based on and why it’s important to understand GCP prices.
This is what we’ll be going through:
What is GCP pricing based on?
Understanding GCP billing
Why it’s important to analyse GCP pricing
Using a GCP pricing calculator
Summary
What is GCP Pricing Based On?
Let’s get a bit of background first. GCP is a suite of cloud computing services that people can use to build and maintain their digital applications. It runs on the Google Cloud infrastructure that Google uses for its own products. This infrastructure is located in different regions and zones across the globe. There are over 20 locations for these data centres.
It’s important to note that Google Cloud is not the same as GCP. Google Cloud is a large collection of digital services, Google’s SaaS offerings,, and includes Google workplace, Gmail and Youtube. Using GCP means you are running your applications on the same infrastructure as the Google Cloud SaaS offerings we all know so well.
GCP offers different services that do different things. These services span across:
Software as a service (SaaS). These are pieces of GCP software that run on the web.
Platform as a service (PaaS). This involves using GCP services to host and run your web applications.
Infrastructure as a service (IaaS). This involves using GCP services to create your own web applications.
Here you can find a list of all of GCP’s products.
How GCP is Charged
When it comes to pricing, web applications that use GCP are charged across two areas - the service(s) used and the amount that the service(s) is used.
GCP offers different pricing models based on the above. On top of this, GCP users can get additional discounts and contracts with different levels of flexibility. Let’s take a closer look at the GCP pricing models.
Pricing Models
GCP offers three different pricing models based on how much their services are used.
Pay-As-You Go
This is known as ‘on demand’. GCP users are charged per minute, per GCP service used. This is the most expensive option, but offers the most flexibility. You can stop and start whenever you want and are only charged for what you use.
Committed Use
This is committing up front that you’ll use GCP service(s) for between one to two years. This type of long term reservation means you’ll get a discount.
Free Tier
You can choose a free tier to test out GCP services and see if they work for your organisation. With this option, you are allocated a predefined set of 24 GCP resources over a set period. These resources all have set monthly limits.
Let’s now look at the different types of services that GCP offers, and how they are charged.
Services
GCP services can be bucketed into three categories: computing/processing power, connectivity/data transfer and storage/memory. These different categories, as well as the resources within them, have different prices.
Computing
GCP offers computing engines. These are PaaS that allow you to run virtual machines across Google Cloud’s infrastructure. A virtual machine is essentially a digital computer. Different engines are optimised to do different things. These can be categorised into accelerator, compute, memory, and general purpose.
These different engines are charged by usage. If you use them for over 25% of a month, you’ll be eligible for a GCP discount of about 20-30%.
Machine Learning
Sitting under the computing category is GCP’s AI and machine learning services; these products are designed to learn from the data you give them.
These are priced by the service being used and then by usage.
Storage
Storage costs are based on what storage functionality you want to use in GCP. These can be split into:
Data storage - the amount and type of data you want to store
Network usage - the amount and type of data that needs to be read or written
Operation usage - the amount and type of storage activity required to manage data
Four types of storage - for data that needs to be accessed more or less frequently. For example, there are specialist storage tiers for archived data. These are cheaper than regular data storage tiers.
You can also be charged for data retrieval, early deletion, and replication of data in different regions.
Big Data and SQL Queries
Big data refers to large amounts of data being collected. Because there’s a lot of it, specialist tools need to be used so it can be analysed, processed, and stored. Big data is often used for machine learning projects. GCP has a range of big data resources. Their prices are all based on storage and usage.
Structured Query Language (SQL) is a computing language used to communicate with databases. GCP offers the Google Cloud SQL database service. This service is priced in relation to:
CPU/memory - based on the region where a VM is running
Storage/networking - based on the region where a VM is running
Shared-core instances - this is a particular type of VM.
Networking
Networking is effectively a form of data transfer. Data transfer can be split into two - outbound (leaving the network) and inbound (coming into the network).
Data transfer costs are based on:
Where the data is
Where it’s going
Other additional services that are needed when transferring it somewhere else.
On GCP, networking costs are separated based on what networking service is being used. Here are a couple of the most common ones.
Virtual Private Cloud
A virtual private cloud (VPC) is a private cloud network set up inside a public cloud. A VPC offers different levels of isolation. This makes the network only accessible to the VPC user.
Divio offers different setups of virtual private clouds. Through Divio, you can have a virtual private cloud service on AWS, GCP, Azure or Exoscale (just reach out if you have a different infrastructure provider you prefer).
For GCP, VPC usage is charged based on traffic. This is the amount of information transferred between networks. There is no charge for traffic coming in, but there are charges for traffic leaving. These types of traffic are charged differently, and based on the following:
Is it going to different boundaries/zones?
Does it have an external or internal IP address?
Is the traffic leaving GC altogether?
How far does the traffic move before leaving GC?
GCP offers two tiers for VPC information transfers. These are premium and standard. ⚠️ But watch out! Premium is set up by default, but might not be the most cost effective for your organisation.
Cloud CDN
Cloud CDN works by caching HTTP or HTTPS load-balanced data near your users. Basically, this can speed up how quickly your application loads. This is particularly important for user experience because it reduces network latency.
Cloud CDNs are priced by:
The number of information transfer requests
The amount of data leaving the network
Filling the cache. The cache is a type of storage layer. It’s where data is stored temporarily. It’s this that allows for quicker application loading speeds.
Understanding GCP Billing
GCP billing is relatively simple. Everything comes through your Cloud Billing console. You can opt for invoiced (offline) billing or self-serve (online) billing. There is a monthly charge cycle, and all payments are automatic.
Why it’s Important to Analyse GCP Pricing
GCP has a pricing calculator you can use to cost up the services you are interested in using.
Whilst there are no termination fees (this may not apply to committed use plans) or upfront costs, GCP does have variable pricing. Like any resource with lots of functionality, you want to make sure you are getting the best bang for your buck.
GCP also has a few hidden data transfer costs. These can include:
Inbound traffic that is processed in a specific way. These include load balancers, Cloud NAT, or Protocol forwarding.
Outbound traffic between VMs in the same cloud zone, where there is no internal IP address.
It’s important to note that costs can rack up if you don’t keep a close eye on your GCP usage. You also want to ensure that you choose the right set of services for you. It’s also best practice to get into good habits. These can include:
Stopping/destroying resources when no longer used
Selecting proper instance storage
Selecting the best storage tiers
Summary
GCP offers long-term discounts, and a relatively clear - as well as competitive - pricing structure. However, it is still important to keep an eye and manage GCP accounts. Similar to other cloud vendors, you always want to make sure you aren’t paying for anything you aren’t using.