Introduction: The cloud has become a normal service, and the selection, flexible delivery, and self-service of cloud resources have become a major trend. Different customers in different industries have different business development stages, and the cost input of cloud resources accounts for different proportions of the overall business cost. Minimizing cost input and maximizing business benefits are the common goals of different customers. Aliyun provides users in the whole industry with a wealth of cloud computing power products and services and flexible sales modes, helping users to accurately predict resource capacity on the cloud and manage resource delivery, which is very helpful for customers to save the cost of purchasing resources on the cloud. This article, the second in the series on private pools on the cloud, focuses on a shopping guide for private pools in different scenarios.
Author: Ali Cloud technology expert Li Yuqian
** Introduction: The cloud has become normal in ** business, and the selection, flexible delivery and self-service of cloud resources have become a major trend. Different customers in different industries have different business development stages, and the cost input of cloud resources accounts for different proportions of the overall business cost. Minimizing cost input and maximizing business benefits are the common goals of different customers.
Aliyun provides users in the whole industry with a wealth of cloud computing power products and services and flexible sales modes, helping users to accurately predict resource capacity on the cloud and manage resource delivery, which is very helpful for customers to save the cost of purchasing resources on the cloud.
This article is the second in the Best Practices — Private Pools on the Cloud series. In the first article, I focused on the value of private pools and how to obtain them. This article focuses on purchasing private pools in different scenarios.
To review what a private pool is: when a user purchases a product such as “Elastic Support” or “Capacity Reservation” on the ECS console, under the “Resource Assurance” service TAB, he or she obtains a pool of deterministic computing resources (CPU and Memory) in the cloud that is exclusively allocated for use. A private pool service has two phases: private pool reservation and private pool resource delivery.
Private pool has the value of resource inventory certainty and resource scheduling delivery flexibility, and can escort the certainty and continuous development of customer business. Choosing the most appropriate private pool for different clients can match resource costs with business growth.
As we know, customers on the cloud come from all walks of life. They realize the value of the industry through digital solutions and digital products and services, and provide various computing services based on the cloud platform. Computing services will eventually be reflected in changes in resource requirements. We abstract the change characteristics of resource demand as shown in Figure 1, which can be divided into three types: daily stability demand, daily elasticity demand and sudden demand.
As shown in Figure 1, the demand of resource certainty is mainly reflected in “daily elastic demand” and “emergent demand”. Among them, “daily elastic” demand can be subdivided into “cyclical” short-term resource demand and “non-cyclical” short-term resource demand (episodic and special period). To sum up, scenarios requiring deterministic delivery focus on:
- “Cyclical” short-term resource needs
- The “occasional” demand for large resources
- “Special time” resource requirements
The following is an introduction to the deterministic resource selection of the three scenarios.
Cyclical short-term resource needs
As shown in Figure 2, resource demand shows obvious periodicity and regularity. The variation characteristics of the number of instances over time are in line with the resource needs of games and online education scenarios. For example, the number of instances increases after work and on weekends, while the number of instances is small on weekdays.
Game scene: XX game opens a new service at a fixed time every Saturday. A large number of users flood in and register, resulting in a surge in resource demand. Online education scenario: XX online education Company, the curriculum of online teaching is concentrated at a fixed time in summer. When the class starts, a large number of resource demands are generated, and the resources can be released after the class ends.
Deterministic delivery plan
In view of periodic short-term resource demand, the following is a classification of resource instance holding time and multi-cloud platform. Under each category, break down the various buying options and show the related advantages and disadvantages.
Plan 1
Hold for a long time
The core of this plan is to buy resources that are periodically needed in advance and at a time. As shown in Table 1:
Scheme 2
Short-term hold
The core of this solution is to ensure that resources are delivered with certainty only when they are needed in a cycle time frame. As shown in Table 2:
In view of the periodic short-term resource demand, it is necessary to charge a certain fee in advance to purchase “elastic protection”. What is the cost input compared with other purchase methods? Let’s do a further analysis.
For example, suppose that the user has a definite resource demand: Beijing region, example specification ECS.g6. Xlarge, 1. The expected cumulative use time is 12 days in a month, and resources can be released in other time periods in a month. Business requirements: When resources are needed at any time, they must be delivered 100% successfully. In this case, to ensure the deterministic delivery of resources, users have four purchase and corresponding charging methods, as shown in Table 3 below:
Method 1: “monthly” of one month, total cost 1 = the monthly price of the instance * 1
Option 2: One month of the “Package year”,
Total cost 2 = average monthly price of the instance package year * Duration (the number of months in this case is 1)
Method 3:0 prepaid RI reserved one month,
Total cost 3 = one month’s RI cost for this instance
Method 4: The elastic guarantee is reserved for one month, and the total cost is 4 = the predetermined cost of the guarantee package + the cost generated during the actual opening period of the instance (40% * 30 = 12 days, the case will be used only 40% of the time in one month). If you buy directly on a monthly basis, you need to pay one month’s price.
Note: A sample cost comparison for determining specifications is as follows:
Monthly average price of package year < monthly price of package month < accumulated monthly price of quantity
Therefore, under the same configuration conditions, the cost relationship of the four different purchase modes is as follows: Total cost 1 > Total cost 2 > Total cost 3> Total cost 4
At this time, with the same configuration, the total hourly cost by volume enabled by “Elastic protection +12 days” is optimal.
The relationship between holding time and cost in the four ways is abstracted as shown in Figure 3:
Note: The price here is a schematic, as the final price will vary depending on the configuration of the instance, and the price of the instance will be adjusted as the market changes. Diagram 3 above mainly expresses “comparative information” : when the usage time is less than 40%, elastic protection is relatively cost-saving. At more than 40 percent, flexi-coverage is no longer cost-effective compared to a monthly plan.
Plan 3
Mixed cycle holding
This program is characterized by a fine-grained management of resources by the business. Big data processing, for example, fixed purchase example, every evening or big promoting activities during buying large quantities of examples of these scenarios is task force showed “level” : the main task is to strictly limit delivery time (need) to ensure resource, the main task can be delayed delivery (dependent on the Spot for instance resources to handle). The alternative purchase options are shown in Table 4:
Plan 4
It’s cloudy combination
The characteristics of this scheme are as follows: when a specific commodity is temporarily cut off on one cloud platform, it is transferred to other cloud platforms for instance purchase.
Occasional large resource demands
Occasional large resource requirements are unpredictable and sudden, as shown in Figure 4. For example, for customers of the Internet or To C service, some hot events trigger a large number of service requests, and a large number of resources are needed To deal with them temporarily.
Deterministic delivery plan
Since it is an occasional large demand for resources, one possible solution is insurance services. Alibaba Cloud’s flexible guarantee is a product and service that has emerged in the context of COVID-19. Users only need to pay a certain amount of money in advance (the cost is related to the amount of resource demands, guarantee time and times of opening), and then, in accordance with the agreement, users can obtain deterministic resource delivery when they have unexpected resource demands. Relevant purchase schemes are shown in Table 6.
Resource requirements at particular times
As figure 5 shows, resource requirements at particular times are planned and predictable. For example, double 11, Spring Festival, statutory long holidays, etc., resources are needed to ensure orderly service.
Deterministic delivery plan
It is generally not recommended to purchase resources for a long time due to short-term resource needs, which is not cost-effective to pay long-term costs. Table 7 shows the relevant optional purchase schemes.
Combination is recommended
The above focuses on analyzing multiple purchase options from a single perspective of the “deterministic” resource delivery component. The “overall” resource demand (business load) of actual customers also shows certain characteristics, as shown in Figure 6. We abstract it into four types: resonance type, smooth type, spike type and mixed type.
For the four classification features of business load, the corresponding purchase recommendations are shown in Figure 7:
conclusion
Based on the resource delivery demands “certainty” scenario, ali cloud provides a variety of choose and buy, such as elastic support, effective immediately capacity, delay effect capacity such as products and services, especially such as electricity, double 11 and 618 other predictable short-term resource requirements of the scenario, effective elastic support or delay capacity reservation is the preferred solution. In the case of periodic resource demand scenarios that require deterministic delivery, immediate capacity reservation combined with SavingPlan SavingPlan is the preferred solution. With deterministic resource delivery, other work such as fine management of resources can be laid a foundation. For more knowledge about cluster resources, you can refer to the book “Deep Clustering – Resource Scheduling and Management in Large Data Centers”.
That’s the end of this best practices sharing and the third in our on-cloud private Pool series — How to plan and implement post-cloud capacity for my business
The original link
This article is the original content of Aliyun and shall not be reproduced without permission.