Summary: This is the final chapter of the private pool on the cloud series. This article will focus on the various solutions for building private pools on the cloud (virtual IDCs), their advantages and disadvantages, and give relevant suggestions for cost-effective optimization.

Author: Ali Cloud technology expert Li Yuqian

Abstract

Around the private pool (virtual IDC) value, access, purchase, capacity planning, the ground to build these aspects, we will introduce one by one in the form of special articles. For example, it introduces the value of private pools and how to obtain them from the perspective of business certainty and continuity. Then from the characteristics of Workload, combined with a variety of demand characteristics, the corresponding purchase scheme suggestions; After learning about the value, acquisition, and purchase of a private pool, you need to know the capacity of a resource before purchasing a private pool. This section describes how to plan and implement the capacity of a resource on the cloud. In this article, we focus on the various options for building private pools (virtual IDCs) on the cloud and the pros and cons of each.

01 Who needs a private pool on the cloud

The core features of private pools include certainty of resource delivery, guarantee of business continuity, exclusive use, and flexible resource sharing by business parties. In addition, the private pool refers to the resource pool on the cloud, which corresponds to the resource pool of the self-built IDC. Figure 1 shows a rough breakdown of customer service choices. Building private pools on the cloud is suitable for enterprise customers.

For enterprise clients, private pools on the cloud are ideal if the resource delivery requirements of the business are certain and the core business continuity needs to be guaranteed. Other super customers, with the ability to build their own machine rooms, also recommend building private pools on the cloud. Cause analysis is shown in Table 1.

Figure 1- Rough classification of customer service options

Table 1 shows the comparative analysis of private pools constructed by traditional self-built IDC and cloud-based OnCloud.

Table 1- Comparison between self-built IDC and OnCloud

For more information

[1] How to build an IDC room?

www.zhihu.com/question/19…

[2] Ding Changyan, How does Huawei accelerate the green development of IDC industry by implementing the goal of “dual carbon”?

T.cj.sina.com.cn/articles/vi…

[3] Jiao Yi Circle, IDC — Industry chain structure and cost structure of data center industry

www.weibo.com/ttarticle/p…

[4] China IDC Industry Analysis Report 2021 – Market Status and Development Opportunities

Baogao.chinabaogao.com/hulianwang/…

02 How do I Create a Private Pool on the Cloud

In the context of distributed cloud, private pools are also distributed.

As shown in Figure 2, the overall layout of a private pool can also take many forms. Different customers eventually have their own ‘private computing power’ in different ‘locations’ such as center, edge, local, field, etc. The sharing scale of center and edge is larger, while the sharing scale of local and field is smaller.

Figure 2- Private pool layout

On the cloud platform, there are currently products and services at both the center and the edge that support customer ‘private pool’ resource delivery. For example, Ali Cloud provides flexible protection, immediately effective capacity reservation, storage exclusive cluster, cloud exhibition, cloud box and other optional commodities and services.

In computing, combining multiple payment models, DedicatedHost, ECS server, ECS server by volume +SavingPlan, elastic guarantee +SavingPlan, and immediate capacity reservation +SavingPlan can be selected to support the construction of cloud private pools of computing resources. The following are analyzed one by one:

Build a private pool based on DedicatedHost

  1. Assess the capacity requirements of the business on resources
  2. The resource capacity requirement is translated into the DedicatedHost quantity requirement
  3. Purchasing a DedicatedHost server (by volume or by year or month)
  4. Create instance resources based on DedicatedHost, such as oversold deployments, custom spec deployments, and so on

This mode is simple to understand: Purchase physical machines on the cloud in batches (no need to pay attention to the operation and maintenance control of physical machines and equipment room, and use physical servers through commercialized, service-oriented and online APIS), and independently arrange application resources based on these physical machines. These services include service load balancing, mixed service load deployment, service load oversold, and time-sharing of resources.

This model is suitable for large enterprises, which have their own resource delivery, resource orchestration, resource management optimization teams, and practice programmable infrastructure. The original self-built machine room, power supply, security, server network management and other supporting tasks were all handed over to the cloud platform. The business side focused on computing power resources and business development to better support business iteration. The business side makes full use of the internal application portrait to deeply analyze the data and build the orchestration, operation and maintenance system matching the application workload characteristics to effectively and controllable improve the resource utilization rate and save costs. For example, XX Technology Group purchased a large number of DedicatedHosts on Aliyun, and built cloud clusters based on these dedicatedHosts to achieve independent secondary scheduling and resource utilization improvement.

2. Build private pool based on ECS server package

  1. Assess the capacity requirements of the business on resources
  2. Purchase and business matching ECS instance specifications, quantity, package year package month long hold

This pattern is simple to understand: A specific resource instance is held for a long time and is not released to support the deployment and orchestration of a specific resource instance.

This model is very adaptable and can meet resource needs of all sizes. In addition, these resources can be allocated to the PaaS platform for capacity-based scheduling and deployment. In this way, the underlying annual and monthly package resources can be shared. This mode requires long-term resource holding and is suitable for scenarios with stable and long-term business development, but not for scenarios with sudden resource demands.

3. Construct private pool based on ECS server quantity +SavingPlan combination

  1. Assess the capacity requirements of the business on resources
  2. Purchase and business matching ECS instance specifications, quantity, quantity of long-term hold
  3. Purchase a SavingPlan that matches the number of resource instance specifications

This pattern is simple to understand: A specific resource instance is held for a long time and is not released to support the deployment and orchestration of a specific resource instance.

This model is very similar to the annual and monthly package, which is to build a “private pool” by relying on fixed retained resources through long-term holding in exchange for preferential purchase discounts. This model is the mainstream cloud vendors to promote the model. In this model, resources and costs are decoupled, and the characteristics of flexibility and low cost are combined.

At the same time, if there is a requirement for the certainty of resources, it can also be paired with immediately effective capacity reservation to ensure the certainty, flexibility and low cost.

4. Combination of elastic guarantee and SavingPlan to construct private pool

The private pools built earlier have one potential “convention” : long-hold scenarios; However, elastic guarantee is more suitable for short-term private resource and deterministic resource delivery scenarios.

  1. Evaluate service periodicality and service duration within a period (for example, the cumulative duration within a month is less than 40%)
  2. Evaluate the specification and quantity of the instance for the periodic part
  3. Purchase Elastic protection (open resource instances at any time)
  4. Purchase SavingPlan (optional)

The pattern is simple to understand: subscribe to a private pool on the cloud, turn resources on when they are needed, and release them when they are not. The reservation fee is charged in one time, and users can control the timing of resource opening.

This pattern matches a business scenario that starts periodically, but has the disadvantage of incurring a certain resource reservation fee (a one-time charge). Currently, the monthly usage time of resources accounts for less than 40%. On the premise of ensuring the certainty delivery of resources, the cost is favorable compared with annual and monthly cost.

5. Effective immediately Capacity reservation +SavingPlan to build a private pool

This model is more flexible and cost optimized.

  1. Assess the resource requirements of the business

  2. Purchase ready – to – release capacity with immediate effect

  3. Purchase a SavingPlan that matches the amount of resources

The pattern is simple to understand: when resource certainty is required, a private pool is booked from the cloud and does not need to be released at any time. Computing resources are charged by volume only when there is free capacity in the private pool. When the capacity is used up, there is no additional cost. When the available capacity is not needed, you can reduce the capacity.

The best practice for this model is to purchase a SavingPlan and book an immediate capacity reservation to gain both cost flexibility and resource certainty.

03 How to improve the cost performance of private pools

To improve the cost performance of private pools, essentially:

  1. In terms of purchasing methods, it can obtain relatively favorable discounts through long-term holding;
  2. In terms of resource utilization, time-sharing multiplexing improves resource utilization.

In view of several private pool construction methods introduced above, the corresponding cost-performance optimization measures are given as shown in Table 2

Table 2- Private Pool cost performance optimization

conclusion

On the cloud, customers need to ensure the continuity of their business services, which requires certainty in the delivery of cloud computing resources. Combining with a variety of purchase modes, Ali Cloud provides’ private pool ‘products and services for different industries and Workload responsible characteristics, supporting users to customize their own private pool on the cloud in key elements such as flexibility, certainty and cost control of resource delivery and sharing.

The resource security service collection of Aliyun (elastic guarantee, immediate effective capacity reservation, annual and monthly instance standby purchase, annual and monthly large-scale resource reporting, large-scale resource reporting by quantity, personalized resource purchase scheme recommendation, etc.) These services are delivered with certainty of resources such as short cycle, long cycle, obvious law, no obvious law, small amount and large Workload.

The original link

This article is the original content of Aliyun and shall not be reproduced without permission.