Google Cloud Machine Types Comparison

Google Cloud Platform offers a range of machine types optimized to meet various needs. A virtual machine can be configured to access a variety of virtual hardware resources, such as a virtual CPU, a virtual disk, and memory. There are a number of types of machines within a machine family that offer different combinations of memory and processor configuration. It can be difficult to choose the right Google Cloud machine for your workload since there are several types available. 



Google Cloud Machine Types Comparison

We are doing an overview of each GCP machine type since we have already covered EC2 instance types and Azure VMs. Here is a quick overview of what we will discuss, but remember that you should research further to determine which machine is the best match for your particular situation.

Google Cloud Machine Type Family

Google Compute Engine manages resources for general-purpose machines. The general-purpose machine type family includes a number of different machine types curated for specific workloads. There are a variety of machine types that can handle a variety of common workloads. Development and testing environments, databases, mobile gaming, and web applications represent some examples of these workloads. High performance and low price make these machines a popular choice. In the General-purpose family, there are four types of machines available: E2, N2, N2D, and N1.

E2 GCM VM  Machine Types

Across all general-purpose machine types, E2 VMs provide a variety of computing resources at the lowest on-demand price. Additionally, E2 machine types offer dynamic resource management, which provides a number of benefits for workloads that prioritize cost savings. You'll see savings of up to 31% over N1 when you use these machine types on Google Cloud. Additionally, E2 VMs already include sustained use discounts and are also eligible for committed use discounts, which can increase savings potential by up to 55%.

This makes it an ideal fit for workloads such as small-to-medium databases, web services, and application development and testing environments that don't demand large instances, GPUs, or SSDs.

N2 GCM VM  Machine Types

There are two types of N2 machines: general-purpose and virtualized. They can support up to 80 virtual CPUs and 640 GB of memory. Using N2 VMs, you will be able to get a 30% increase in performance from your virtual machines, as well as shorten many of your computing tasks. For VMs created with the extended memory feature, these types of machines offer higher memory-to-core ratios. 

Applications, web and application servers, gaming servers, content management, and collaboration systems are all good candidates for general-purpose workloads.

N2D GCM VM Machine Types

With up to 224 vCPUs and 896 GB of memory, N2D machines are the most powerful general-purpose machines. You can use these VMs in the same way as N2 VMs.Streaming video, web applications, databases, and workloads are a good fit.

N1 GCM VM  Machine Types

The N1 virtual machines support up to 96 vCPUs and 624GB of memory and are first-generation general-purpose virtual machines. The N1 virtual machines do offer a larger sustained use discount than N2 virtual machines, despite the recommendation to use one of the second-generation general-purpose machines. Additionally, they are capable of supporting Tensor Processing Units (TPUs) in some areas. 

Compute-Optimized Machine Type Family

Compute-intensive workloads are best handled by computer types optimized for computing. Compute Engine's core-per-core performance is the highest and most consistent for these machine types. Game servers, API server workloads with a high latency requirement, and HPC workloads are ideal for computer-optimized machines. These machines have 40% greater performance than previous generation N1. 

C2 Machine Types

 C2 Machine Types is optimized for computing are ideal for computing-intensive workloads. Compute Engine's most consistent performance and highest performance per core is offered by these machine types. Dedicated computer types can be used for game servers, latency-sensitive API servers, and high-performance computing (HPC). Compared to N1 machines of the previous generation, these machines have 40% better performance.

Memory-Optimized Machine Type Family

In addition to offering the highest memory configurations across every VM family, these machine types offer the highest memory-to-vCPU ratios with a maximum of 12 TB for a single instance, making them the ideal choice for tasks requiring intensive memory usage. There are three main types of machines that do not support GPUs. Memory-optimized machines offer sustained use discounts of up to 30%. Further, they may also qualify for discounts for committed use, bringing the total savings to over 60%. For in-memory databases and in-memory analytics, these are the most suitable machines. 

M2 Machine Types

M2 VMs support the most demanding and business-critical database applications with up to 12TB of memory. Compute Engine offers these machine types at the lowest per-GB price, making them a good choice for workloads that use higher memory configurations but have low computing needs.

In July 2020, accelerator-optimized virtual machines were added. In this type of machine family, demanding computation workloads are optimized - these include high-performance computing and CUDA-enabled machine learning.

A2 Machine Types

In comparison with previous-generation NVIDIA V100 GPUs, every A2 VM is powered by A100 GPUs that offer 20x improvements in computing speed. Google is currently making these machine types available through its alpha program.  

Shared-Core Machine Types

Shared-core machines are a cost-effective choice for workloads that run for a short period of time, such as small batch jobs. Partially virtualized processors are based on a hyper-thread running on the host CPU that runs your instance. By using context-switching, these types of machines can multitask on a single physical core by sharing it among their vCPUs. 

The GCP family of shared-core processors can deliver large blocks of physical CPU time for short periods when needed. These are spikes in compute power that usually occur when your workload requires more power than you had allotted. These bursts are not permanent and only occur periodically.

Custom Machine Types

These predefined types of machines can cater to needs based on memory, vCPU, a balance between both, or both memory and vCPU. In the event that none of the available machine types meet your needs, Google offers customers the option of customizing their machines. You can define exactly how many virtual CPUs and how much system memory an instance needs with custom machine types. So you're only paying for what you use, you can balance CPU and memory independently. Choosing them is a great solution when your workloads don't quite match up with any of the predefined types, or if you need more compute power or more memory without getting bogged down with upgrades you don't need.

What Google Cloud Machine Type should you use?

The Google Cloud offers enough variety for almost any application between the predefined options and the ability to create custom machine types. When it comes to pricing, the machine you choose matters less than the cost because the resource-based pricing structure takes your inputs into account.

When you have a clear understanding of your workload and usage trends, you will be able to choose a machine type that fits your business needs.

About GPUs and machine types

Google Cloud Platform also offers graphics processing units (GPUs), which can be used for processes such as machine learning and data processing. You can only attach GPUs to predefined or custom machine types. Generally, the more GPUs you have attached to your instances, the more vCPUs and system memory you have available.

Post a Comment

0 Comments