圖片來源：https://cloud.google.com/about/locations/?hl=zh_TW#regions

With the development of the cloud industry, more and more enterprises are introducing cloud services in their digital transformation process or business expansion. Cloud services can help companies seamlessly and quickly connect their businesses in a short period. Therefore, it is necessary for enterprises to have a deep understanding of cloud products and know which products can help the company grow.

Google launched Google Cloud Platform (GCP) in April 2008. The image above shows the locations of Google Cloud's 23 data centers worldwide, including one located in the Changhua County of Taiwan's Changbin Industrial Zone, which was inaugurated in December 2013. In recent years, Google has formally announced plans to build two new data centers in Tainan and Yunlin.

The services and products provided by Google in the cloud can be accessed and utilized through the GCP platform. Individual users can use their Google accounts and bind their credit cards to the GCP platform to start using GCP services. Enterprise users are advised to contact trusted distributors to provide more affordable prices. Haier Cloud also provides services on the GCP platform for customers. Please feel free to contact us.

So far, there are over a hundred services on GCP, such as Compute Engine for computing, Storage for cloud storage, Load Balancer for cloud load balancing, BigQuery for data analysis, and more. How should we understand cloud products, and where should we start? In the coming weeks, we will select the most commonly used services on GCP in the industry for more in-depth introduction and explanation. We hope that through our column, you will have a better understanding of cloud service products.

Source：https://raw.githubusercontent.com/gregsramblings/google-cloud-4-words/master/DarkPoster-lowres.png

This week, let's start by introducing Google Compute Engine, or GCE for short. Before delving into GCE, let's take a look at the chart above. This chart was created by the Google Developer Relations Team in March 2021 and provides an overview of the products. We can see that there are a total of 20 product categories in the chart. When initially using cloud services, it's not necessary to be familiar with every item on the chart. Instead, it's important to find the products that best meet your needs. If you haven't used cloud services before, we recommend starting with the major categories: Compute, Storage, Networking, and Identity and Security, to gain an understanding. And today, we'll be introducing GCE, which belongs to the Compute category.

What is GCE?

Google Compute Engine, abbreviated as GCE, is a Infrastructure as a Service (IaaS) offering belonging to the compute category. In simple terms, GCE provides virtual machines (VMs). Users can design a virtual machine tailored to their specific business needs and launch it quickly to meet those requirements.

Introduction to Machine Configuration

Before using a GCE virtual machine, users must clarify what tasks they intend to accomplish with it. Are they planning to run a static website, perform data analysis, or engage in extensive machine learning? Since different machine types incur different pricing, having a clear usage goal helps select a machine configuration that not only meets requirements but also saves costs, achieving a win-win situation.

The utility of "Machine Families" and "Machine Types" on GCE is to choose the most suitable feature orientation, vCPU, and memory for the virtual machine. It is divided into two parts: users confirm the machine usage scenario (Machine Families) and then select the specifications (Machine Types).

1. Machine Families

• General-purpose
• Compute-optimized
• Memory-optimized
• GPU

Their detailed functionalities are as follows:

General-purpose

The first category is general-purpose, suitable for typical workloads, and includes models such as E2, N1, N2, and N2D. These models can satisfy the majority of general usage scenarios. To delve deeper, within the general-purpose category, there are cost-optimized E2 and balanced N-series.

E2 is characterized by cost optimization, aiming to provide everyday services at lower costs. Suitable services for E2 include web serving, Line-Of-Business apps (e.g., e-commerce systems, workflow management systems, file management systems, development/test environments, microservices), and medium-sized databases such as Access, FoxBASE, MySQL, and SQL Server.

On the other hand, N1, N2, and N2D are suitable for enterprise applications (ERP, CRM, SCM, HRM, CMS, and BI), medium-sized databases (SQL Server, MySQL, Informix), web and app serving, media services, etc.

E2, N1, N2, N2D machine types come pre-configured with vCPU and Memory, but the ratios differ. For instance, E2 has a vCPU-to-memory ratio of 2:8, N1's ratio is 1:3.75, and N2 and N2D have a ratio of 2:8. N2 supports the Intel Cascade Lake CPU platform, while N2D supports the second-generation AMD EPYC Rome platform.

Compute-optimized

The second category, Compute-optimized, is designed for high-performance computing tasks and includes the C2 model. C2 emphasizes compute optimization and is suitable for services such as Electronic Design Automation (EDA), High Performance Computing (HPC), Scientific modeling, and AAA Gaming.

Memory-optimized

The third category, Memory-optimized, is for tasks requiring large memory capacity and includes the M1 model. M1 emphasizes memory optimization and is suitable for services such as SAP HANA (a commercial application database for predictive analytics), real-time data analytics, in-memory cache, and in-memory databases.

GPU

The fourth category, GPU, is for high-performance computing, visualization, and machine learning tasks and includes the A2 model. GPUs excel in processing images, graphics, audio, and graphics. A2 is optimized for accelerators and is suitable for services such as High Performance Computing (HPC), Machine Learning, and parallel computing. It's worth noting that A2 is currently only available in specific regions, and it's not available in the Taiwan region.

2. Machine Type

2. Once the machine family is determined, the next step is to select the vCPU and memory in the "Machine Type." GCE provides default options and also allows users to customize their settings. You can refer to the following chart for more details.

The choice of vCPU and memory in GCE is highly flexible. For instance, in the E2 "standard" machine type, there are five default options available, ranging from 2 to 32 vCPUs and 8 to 128 GB of memory. Users can also customize these settings, with vCPU quantities ranging from 2 to 32 and memory from 1 to 16 GB.

Boot Disk 

Once the specifications of the virtual machine are determined, the next step is to select the operating system and boot disk size. GCE offers four options: Public Images, Custom Images, Snapshots, and Existing Disks. Depending on your company's business needs, you can use one of the 14 operating systems available in the platform's public images, including popular ones like CentOS, Debian, Red Hat Enterprise Linux, Red Hat Enterprise Linux for SAP, SQL Server on Windows Server, Ubuntu, Ubuntu Pro, Windows Server, etc. Alternatively, using custom images created by your company is also an option.

The choice of vCPU and memory in GCE is highly flexible. For instance, in the E2 "standard" machine type, there are five default options available, ranging from 2 to 32 vCPUs and 8 to 128 GB of memory. Users can also customize these settings, with vCPU quantities ranging from 2 to 32 and memory from 1 to 16 GB.

Boot Disk Once the specifications of the virtual machine are determined, the next step is to select the operating system and boot disk size. GCE offers four options: Public Images, Custom Images, Snapshots, and Existing Disks. Depending on your company's business needs, you can use one of the 14 operating systems available in the platform's public images, including popular ones like CentOS, Debian, Red Hat Enterprise Linux, Red Hat Enterprise Linux for SAP, SQL Server on Windows Server, Ubuntu, Ubuntu Pro, Windows Server, etc. Alternatively, using custom images created by your company is also an option.

As for boot disk types, there are four options available, with a maximum limit of 65,536 GB each:

1. Standard Persistent Disk: Using HDD drives.
2. Balanced Persistent Disk: Using SSD drives, stable and cost-effective, which is the default option.
3. SSD Persistent Disk: Using SSD drives.
4. Extreme Persistent Disk: Using SSD drives, suitable for high-performance workloads with high IOPS.

Summary

This week, we introduced Google Cloud's operating platform - GCP, and mentioned its data center location in Changhua, Taiwan, with plans to build two new data centers in Tainan and Yunlin in the future. Google Cloud offers over a hundred services categorized into 20 major items. We recommend users who have not yet engaged with cloud products to start with computing, storage, networking, identity permissions, and other key areas for understanding.

Next, we introduced GCE under the Compute service, which allows users to deploy virtual machines. We suggest users clarify their usage scenarios first and then choose the "Machine Families" accordingly. We also introduced four machine families and outlined the suitable business scenarios for each. Additionally, we discussed how to select and configure vCPUs and memory for virtual machines in the "Machine Type" section. Lastly, we provided an overview of the boot disk options available on GCE, including default operating system choices and disk sizes.

We believe that through this introduction, you have gained a deeper understanding of GCE, and this series of articles will continue to explore the features of GCP. Next week, in "Exploring GCP with Haier - Cloud Storage," we will bring you a wealth of knowledge about storage features. Please stay tuned for our column articles. See you next week!