Cloud Dry Goods | Essential for Cost Reduction—Basic Principles of Elastic Scaling

Auto Scaling is a service provided by cloud providers that automatically adjusts computing capabilities (ie, the number of instances) according to their own business needs. When using this service, the number of instances can be automatically scaled according to the workload, so as to provide resource support when business needs grow, and reduce costs when business needs drop.

The SpotMax cloud resource optimization service and the MaxGroup tool in it can be fully integrated with the elastic scaling function to maximize the use of Spot instances under the premise of ensuring the stability of user services, thereby reducing cloud costs. link to learn about 161d2c9de245ce SpotMax

So, what is the basic principle of elastic scaling? How do we judge the effect of elastic scaling and optimize it? Let's find out——

one. What is elastic scaling
Resilience-the server should have a certain tolerance for changes in business volume;

Scaling-In response to changes in business volume, the system needs to adjust resources online. After the expansion is completed, the external service capability of the system should be restored to a normal level.

We need to use elastic scaling to achieve 2 major goals: 1. Reliable service; 2. Resource cost saving

First, understand the deployment architecture of the elastic scaling service:

After the request comes in from the gateway service (ie LoadBalance), it reaches the application (App). The cloud host required by the application is often handed over to the scaling group to manage. The scaling group will automatically purchase new machines and recycle old machines according to the usage of host resources, so as to keep the resource usage of the entire cluster at the expected level.

So, how does the telescopic group complete this series of "cool" operations?

"Expansion" is mainly concentrated in two directions:

1. In the horizontal direction, the scaling effect can be achieved by adjusting the number of nodes by adding and subtracting nodes.
2. Vertical direction: through adding configuration and subtracting configuration, to achieve the effect of expansion and contraction.

In addition, if we want to be flexible, we must monitor the use of resources. As shown in the flowchart, the indicator monitoring service is responsible for real-time data collection from the APP and the cluster, and the collected results will be handed over to the autoscaler (scalable scheduler). Autoscaler will make a judgment based on a predetermined strategy (for example, based on CPU or QPS), and may set a scaling threshold. If it is judged that the expansion or contraction is triggered, Autoscaler will contact the node group to purchase machines or recycle machines. The nodes after scaling will be synchronized with the gateway, thus completing a basic elastic scaling web service.

two. What is an excellent elastic scaling solution?

Each step of the elastic scaling process affects the effect of elastic scaling in different ways:

First, , monitoring service as the starting point of elastic scaling, its data validity and real-time nature will affect the accuracy and sensitivity of scaling; and the richness of monitoring indicators determines the operational range of the strategy;

followed by , Autoscaler's algorithm and strategy directly determine the effect of the scaling strategy.

again , the strength and speed of expansion will directly affect the efficiency of expansion and the utilization rate of resources.

finally , the application as the carrier of the service itself, its pull speed, startup speed, and startup success rate directly determine the final result.

In summary, an excellent elastic scaling solution should try to meet the three elements.

First, it is reliable, it must have a certain resistance to sudden pressure increase
Second, it is efficient. It can quickly adjust to the expected resource usage.
Third, economy. It must increase the utilization rate of resources as much as possible within a reasonable range and reduce waste.

[Elastic scaling and cost reduction partner: SpotMax cloud resource optimization service]

For cloud-using companies, as their business grows, cloud-using costs will increase rapidly, becoming a "stumbling block" that restricts their rapid development. SpotMax as a cloud resource optimization service , of which MaxGroup tool flexible features of the service instance to help enterprises to use the elastic and scalable SpotMax to solve the problem of insufficient resources. At the same time, reduce the cost of using the cloud.

Taking AWS as an example, under normal circumstances, when the Spot instance is insufficient, users cannot continue to apply for instance resources when the AWS Auto Scaling service is insufficient, resulting in a long-term shortage of resources and users facing the risk of heavy traffic impact. When MaxGroup finds that autoscaling cannot apply for resources, it can promptly supplement users with on-demand instances to fill the resources required by the service. When it can continue to apply for Spot resources, MaxGroup will actively replace on-demand instances with Spot. Under the premise of stable service, the goal of guaranteeing the lowest cost is guaranteed. Long-term practice has proved that SpotMax can help enterprises save 60% of cloud usage costs on average, and can reduce costs by 90% at most.

In the cloud, when resources can be scaled on demand, the only limit you can imagine is cost. The SpotMax solution can help companies achieve extreme cloud cost savings and release their business growth potential. Want to learn more? Click the "link below" to get your cloud cost saving cheats with one click!

Understanding SpotMax
Learn about MaxGroup