What Is GCP Compute Pricing?
Google Compute Engine is a cloud computing service provided by Google that allows users to run virtual machines on the Google Cloud Platform.
The pricing for Google Compute Engine is based on the type and number of virtual machine instances you choose to run, as well as the amount of memory and storage that you need. Google also offers various discounts and pricing promotions, such as sustained use discounts for instances that are used consistently over a longer period of time, and committed use discounts for instances that are reserved in advance.
Additionally, Google Compute Engine costs may vary depending on the region in which the instances are run and the specific features and services that are used.
In this article:
GCP Compute Instance Pricing
General-Purpose Machine Types
Google Compute Engine offers several different machine types that are designed for different types of workloads. General-purpose machine types are designed for workloads such as web servers, small databases, and development environments. These machine types are typically a good starting point for users who are new to Google Compute Engine, or for users who have workloads that don’t require the specialized features of other machine types.
An example is the E2 series of machine types. These machine types are designed for use cases like real-time data analysis, scientific computing, and large databases.
Some examples of E2 machine types are:
- e2-standard—this is the basic E2 machine type, with a balanced ratio of CPU to memory.
- e2-highmem—this machine type is suitable for workloads that require a lot of memory.
- e2-highcpu—this machine type is well-suited for workloads that require a lot of CPU power.
The table below shows pricing examples for E2 machine types.
Compute-Optimized Machine Types
Google’s compute-optimized machine types are designed for workloads that require a high amount of CPU power, such as batch processing, video rendering, and scientific simulations. These machine types provide a higher ratio of CPU to memory.
Google Compute Engine’s compute-optimized machine types use Non-Uniform Memory Access (NUMA) architecture, which can improve the performance of certain workloads that are sensitive to memory access latency. In a NUMA architecture, each processor has its own local memory, and processors can access the memory of other processors, but this access is slower than access to their own local memory.
Some examples of compute-optimized machine types offered by Google Compute Engine are:
- c2-standard—this is the basic compute-optimized machine type, with a higher ratio of CPU to memory.
- c2-highcpu—this machine type has even more CPU power than the standard machine type.
The table below shows pricing examples for C2 machine types.
Memory-Optimized Machine Types
Google’s memory-optimized machine types are designed for workloads that require a high amount of memory, such as in-memory databases, real-time data analysis, and scientific computing. These machine types provide a higher ratio of memory to CPU than other machine types.
Some examples of memory-optimized machine types offered by Google Compute Engine are:
- m2-standard—this is the basic memory-optimized machine type, with a higher ratio of memory to CPU compared to compute optimized machine types.
- m2-highmem—this machine type has even more memory compared to m2-standard
- m2-megamem and m2-ultramem—these machine types have the most memory of any Google Compute Engine machine type.
The table below shows pricing examples for M2 machine types.
Accelerator-Optimized Machine Types
Google Compute Engine offers several different machine types that are designed for different types of workloads. The accelerator-optimized machine types are designed to provide the best performance for workloads that can benefit from the use of accelerators, such as graphical processing units (GPUs) or tensor processing units (TPUs).
Some Google Compute Engine’s accelerator-optimized machine types provide the NVIDIA A100 GPU accelerator, released by NVIDIA in 2020. It is designed for use in machine learning, scientific computing, and other demanding workloads. The A100 GPU offers significantly more performance than previous NVIDIA GPUs.
Some examples of accelerator-optimized machine types offered by Google Compute Engine are:
- n1-standard-4-gpus: This machine type has four NVIDIA Tesla V100 GPUs, which are well-suited for machine learning and other workloads that require a lot of graphical processing power.
- n1-highmem-8-tpu: This machine type has eight Google TPU v3 pods, which are well-suited for machine learning and other workloads that require a lot of tensor processing power.
Pricing example for E2 shared-core machine types:
3 GCP Compute Engine Pricing Factors
Billing for Google Compute Engine is based on the type and number of virtual machine instances you choose to run, as well as the amount of memory and storage that you need. Google charges for Compute Engine instances on a per-second basis, with a one-minute minimum usage.
Instance Uptime
Instance uptime refers to the amount of time that an instance is running. Google Compute Engine charges for instances based on the amount of time they are running, so instances that are running for longer periods of time will incur higher charges.
Instance states refer to the different states that a Compute Engine instance can be in, such as running, stopped, or terminated. The charges for an instance will vary depending on its state. For example, instances that are running will incur charges, while instances that are stopped will not incur charges.
Resource-Based Pricing
Google Compute Engine uses resource-based pricing for vCPU and memory, which means that the price of an instance is based on the amount of CPU and memory that it uses. The price of each vCPU and unit of memory varies depending on the machine type and the region in which the instance is run.
Discounts
Google Compute Engine offers various discounts and pricing promotions to help users save money on their cloud computing costs. The specific discounts available and the percentage range of the discounts may vary over time. Currently, the primary discount types are:
- Sustained use discounts—these discounts are available for instances that are used consistently over a longer period of time. The exact percentage of the discount will vary depending on the length of time the instance is used, but the discount can be as high as 30% for instances that are used for a full month.
- Committed use discounts—These discounts are available for instances that are reserved in advance for a specific period of time, such as one or three years. The exact percentage of the discount will vary depending on the length of the commitment, but the discount can be as high as 57% for a three-year commitment.
- Spot VM discounts—these discounts apply for instances that are run on Google’s excess capacity. A spot VM is a type of Compute Engine instance that can be purchased at a discounted price, but it may be terminated by Google at any time if the excess capacity is needed for other purposes. The spot VM discount can be as high as 90% off the regular on-demand price of an instance.