Introduction: After an enterprise accesses the cloud, the budget on the cloud directly affects the priority, progress, and depth of the cloud. The level of budget input is closely related to capacity assessment of business development and resource requirements. Accurate capacity assessment can make cloud budget planning more scientific and more suitable for the needs of business development stage. This article shares how to plan and implement capacity for post-cloud enterprises.
Author: Ali Cloud technology expert Li Yuqian
Abstract
With the digital transformation of enterprises and the rapid development of enterprise IT service cloud, the pace of customers’ going to the cloud is more compact, and the budget on the cloud directly affects the priority, progress and depth of going to the cloud. The size of the budget is related to business development, and another key factor is capacity assessment of resource requirements.
Accurate capacity assessment can make cloud budget planning more scientific and more suitable for the needs of business development stage. This article will share how to plan and implement capacity after cloud service.
First, why to carry out capacity planning
Digital transformation of enterprises, enterprise IT service cloud is making great strides in the development of cloud or cloud enterprises, the regular budget expenditure includes digital information or IT software service expenditure. This part of the budget expenditure, including cloud resources budget input, one of the accounting basis: cloud capacity planning and implementation.
Scenarios that require “capacity” planning are common in everyday life. For example, reservoir water storage is a typical dynamic “capacity” planning process, which needs to be regulated according to the upstream and downstream water environment. For example, during the epidemic period, scenic spots implemented the measure of tourists purchasing tickets after making reservations in advance, and the total number of tourists per day should be regulated according to the prevention and control requirements.
Similarly, services on the cloud will also develop and change dynamically, and the computing power resources on which cloud products and services depend also need to be adjusted accordingly. We abstract the quantity planning of computing power resources into capacity planning.
The necessity of capacity planning is that the services of an enterprise develop dynamically, and the computing power resources on which the services depend also need to be adjusted dynamically. Too much computing power resources lead to idle resources and waste of costs. Too little computing power resources affect service response performance and hinder rapid business development. So, what are the problems if the enterprise goes to the cloud without capacity planning?
First, there may be a mismatch between cost input and business development. Presents the fast development of the situation, for example, when business business depends on work force resources demand also showed a trend of rise, at this point, if there is no capacity planning, probably business outbreak period of time, the back-end service capacity can’t keep up with in time, will affect the business sustained, stable development, even miss the gold business development opportunity.
In addition, the application of Internet technology has greatly shortened the distance between service consumers and service providers, and the cross-regional high availability and stability of service performance of service providers has become a normal goal. To achieve this goal, the most direct solution is to implement capacity redundancy between regions so that traffic can be switched over for DISASTER recovery in the event of hardware or software failures or other emergency scenarios.
To sum up, capacity planning is critical and needs to be continuously planned. Accurate capacity planning can help rapid business development and prevent computing power support from becoming bottlenecks and obstacles for business development. Meanwhile, high availability and stability of cross-regional services can be guaranteed.
Second, business requirements into capacity planning
Capacity planning is for services. Capacity planning is meaningless if it is independent of the actual service status. It is only reasonable to make capacity planning that matches service development according to service characteristics and development goals.
For example, in enterprise A, each employee in department B needs an office computer. Currently, I purchase cloud desktop products of Ali Cloud. As the number of employees in department B is expected to increase by 10% this year, the capacity of cloud desktop should also increase by 10% this year. This example is intuitive and easy to understand. In fact, there are many factors that need to be considered in cloud capacity planning for different industries and different business characteristics. The following disassembly analysis is carried out according to the general understanding, as shown in Figure 1, gradually subdivided from bottom to top.
Factor 1: Overall development assessment of business needs
The overall development trend and assessment of enterprise business is the foundation of all demand sources. Without the full assessment of overall business development, it is impossible to output reasonable and effective capacity planning assessment. For enterprises, capacity planning is not for the sake of capacity planning. Capacity planning is for business development. The overall business development assessment is naturally at the bottom of the pyramid.
Factor 2: Business requirements Development assessment of the native part of the cloud
The next level up from the bottom of the pyramid corresponds to the assessment of the development of the cloud native part. The proportion of the development of cloud native services is directly related to the proportion of the on-cloud capacity planning budget. For the Internet industry, the main body of possible business is cloud native; For traditional industries, if only the enterprise management information part of the cloud, then the development evaluation of the cloud native part is a very small proportion.
Factor 3: Need assessment of priority assurance on cloud under limited budget
For enterprises, the budget of each item is always limited, and the limited resource services should give priority to the development of key businesses, so as to maximize the input-output ratio. For all cloud services, storage, database, and computing services are basic dependencies. Generally, planning and investment of these three components are guaranteed with high priority.
Factor 4: Continuity requirements assessment for the native part of the business cloud
For an enterprise, business continuity is critical at all stages of business development, especially the continuity of key business services. Therefore, the capacity planning process needs to focus on and evaluate the business continuity reflected in the budget. For example, computing resources that core services depend on can be planned to ensure the continuity of services, such as annual and monthly instances, elastic resource guarantee service, and resource reservation service.
Resources: Resource Assurance Services
Help.aliyun.com/document\_d…
Factor 5: Regional Dr Requirement assessment of the service cloud native part
For enterprises, at different stages of development, the priority of services in the region may be different, so capacity planning needs to be aware of the region. Meanwhile, the high availability of services often depends on the construction of inter-regional service disaster recovery capabilities. Therefore, the budget needs to balance the needs of regional development.
Factor 6: Service cloud native partial requirements independent versus comprehensive planning
Based on the first five factors, capacity estimation becomes more specific. Then, starting from factor 6, planning needs to consider the impact of specific operations. Independent planning and comprehensive planning depend on different inputs and output schemes. For example, in an employee office scenario, cloud desktops are independent of each other and can be planned and delivered independently.
For example, large Web services rely on cloud databases, cloud storage, and traffic bandwidth. Therefore, capacity evaluation needs to be packaged and delivered as a whole to avoid short-board effect. And different tools and solutions are relied on to estimate how much capacity is available. For independent programs, general assessments are relatively easy to give; For comprehensive planning, Ali Cloud capacity planning service provides a full set of solutions.
Resources: Capacity planning services
www.aliyun.com/service/cap…
Factor 7: Evaluation of current discount information of different cloud service providers
After the segmentation of business capacity planning is in place and the products and tools on which capacity planning depends are identified, the next step is to perceive discount information.
Different cloud service providers have related activities and discounts in different regions and computing power products. Evaluating this part of the content can make the same budget to buy more affordable computing resources. For example, Aliyun launched SavingPlan + CapacityReservation service, which realized cost saving and deterministic delivery of resources.
Factor 8: Evaluation of planned capacity delivery schedules
The capacity delivery schedule assessment step is to output specific planning information such as when, where, which computing resources to deliver, and the corresponding budget. Delivery too early or too late can be mismatched with business development, or even capacity planning may not be implemented.
Capacity planning is mapped to resource purchases
In the previous section, we described the factors that need to be considered in capacity planning in a hierarchical manner from the bottom up. The essence of planning evaluation is to meet the development needs of business at the appropriate time and place, and to plan the computing power requirements at the corresponding time and place.
As figure 2 shows, there are many ways to map specific requirements to computing power. The following assumption is made: the cloud service capability required by the future development of enterprise business is predictable. Based on the predictable value, it is transformed into the specific purchase demand of resource instances, and then a specific purchase plan is formed. Common technical solutions for mapping planned capacity to resource purchases are described below.
Method 1: linear mapping – horizontal scaling
From a resource perspective, the classic evaluation method is: total resource instances = total business requests QPS/ QPS supported by a single resource instance. When service development requires more computing power, the total QPS changes. In this case, the number of resource instances to be added equals new QPS/ single-node QPS. This method corresponds to what is called “horizontal capacity expansion” in resource scheduling. Aliyun provides services such as Auto Scaling that support automatic horizontal Scaling.
Resources: Elastic expansion
Help.aliyun.com/document\_d…
For more information about Horizontal Pod Autoscaling, see THE HPA (Horizontal Pod Autoscaling) of K8s:
Kubernetes. IO/useful/docs/tas…
Method two: linear mapping – vertical scaling
From a resource perspective, vertical expansion is relative to horizontal expansion. The total number of resource instances is adjusted to adjust the total number of service request QPS by adjusting the computing power of a single resource instance, that is, the QPS supported by a single resource instance (indirectly reducing the QPS supported by a single resource instance through the reduction of resource instance allocation). In the scenario of refined resource scheduling and mixed service load deployment, the capacity of a single resource instance is expanded vertically.
This vertical scaling has two forms: one is fixed (the specifications will not change after adjustment), for example, from the original 4VCPU, vertical scaling to 2VCPU. Then the instance expands and shrinks horizontally by 2VCPU. The other is non-fixed (elastic scaling of a single computing force resource in a short period of time). For example, during the operation of a resource instance, the resource of a certain dimension is “restricted”, so as to adjust the computing force of a single instance resource in a specific scenario.
For the business side, the instance specifications you see have not changed. A typical RESOURCE model such as K8s, such as CPU resource application, has two parameters, request and limit, which can achieve the elastic burst of CPU resources. Another example is ali Cloud burst performance instance. CPU integral is used to ensure computing performance. This example applies to scenarios where CPU usage is low at normal times but occasionally has high CPU usage.
Resources: Burst performance examples
Help.aliyun.com/document\_d…
On vertical scalability, let more content can refer to GKE VPA (vertical – pod – autoscaler) : cloud.google.com/kubernetes-…
Method three: Nonlinear mapping – full link evaluation
Large-scale Internet services, such as e-commerce transaction system, are characterized by multiple business scenarios, inter-business dependencies, and large scale of business services. It is difficult to assess the system capacity individually by application. Therefore, you need to assess the system capacity as a whole in a full-link scenario.
Aliyun’s capacity planning service provides a full set of services, including:
- Service planning, providing business traffic analysis, data capacity analysis, message capacity analysis, database capacity analysis, cluster capacity analysis;
- After service planning and execution, the service provides the full-link pressure test solution, scenario traffic ratio, scheduling solution, traffic limiting and degrading solution, and drill solution.
The core value of full-link evaluation is to help customers detect the optimal pressure, ultimate pressure and failure pressure point of the system on the cloud, and perform degradation and current limiting protection. Full-link evaluation is especially suitable for large-scale and complex scenarios.
Resources: Capacity planning services
www.aliyun.com/service/cap…
Method four: capacity prediction – automatic deployment
Compared with Methods 1, 2 and 3, Method 4 accurately evaluates future capacity changes before work, implements automatic resource delivery based on system load balancing and system QPS water level monitoring, including automatic horizontal expansion, capacity reduction and cross-specification instance delivery, etc. For example, Ali Cloud elastic container instance ECI supports multi-specification instance delivery. The automated o&M service on the cloud provided by OOS can automate the management and execution of tasks. Customers can use templates to define tasks to execute, order to execute, input to execute, and output to execute, and then execute templates to automate the execution of tasks. OOS supports cross-product usage, and you can use OOS to manage cloud products such as ECS, RDS, SLB, VPC, etc.
Resources: Elastic container example
Help.aliyun.com/product/874…
O&m Choreography services
Help.aliyun.com/document\_d…
In summary, the process from business requirements to resource capacity planning to resource capacity execution can be summarized as shown in Figure 3.
Iv. Purchase plan of resources
When the purchase amount of resources is clear, the specific purchase plan is shown in Figure 4. On the time axis of business development, the computing resources are delivered with certainty.
As described in the three Typical Scenarios virtual IDC (Private Pool) Purchase Guide, service resource delivery has the following requirements: Daily stable resource requirements, daily elastic resource requirements, and sudden resource requirements. Enterprises need to choose appropriate resource purchase schemes according to their own business development characteristics and specific resource requirements to achieve cost saving and resource certainty delivery. For example, you can choose products and services related to elastic resource protection for periodic resource requirements, occasional resource requirements, and specific period resource requirements. Detailed capacity planning to the final implementation of the purchase scheme, you can directly refer to the article.
Cloud on the enterprise business, the need for resources, capacity planning, ali cloud provides a rich product power support business capacity accurately evaluate, flexible, the choose and buy, especially the security services based on resources, such as elastic support, effective immediately reserve capacity, etc, certainty delivery support resources, guarantee the continuity of business development.
The original link
This article is the original content of Aliyun and shall not be reproduced without permission.