Mixed Scene 1:
The business experiences noticeable peaks and valleys within a certain period, along with sudden, non-periodic short-term business peaks.
Example:
Every weekday at 8:30 AM, a fixed workload analysis and statistics task is required, lasting for 2 hours, which needs 1 additional node for computing power. At other times, the required computing power for unexpected peak loads is uncertain. In this case, three scaling rules can be configured to ensure sufficient computing power and cost savings.
Rule 1: Set a time-based scale-out rule by choosing to repeat the rule every Monday/Wednesday/Thursday/Friday and schedule it to scale out 1 node at 8:15 AM tomorrow, with a scheduled termination duration of 3 hours.
Rule 2: Set a load-based scale-out rule, and select the monitoring metrics as needed. It is recommended not to set a validity period, with the default being active all day, scaling out 1 node.
Rule 3: Set a load-based scale-in rule, and select the monitoring metrics as needed. It is recommended not to set a validity period, with the default being active all day, scaling in 1 node.
Note:
1. Scaling out resources takes time, and the time required is proportional to the number of resources being scaled out. It is recommended to prepare resources at least 15 minutes in advance. Typically, the time required is relatively short.
2. The priority order for the three scale-out rules is Rule 1 > Rule 2 > Rule 3; the number of resources to scale can be adjusted based on actual business needs.
Mixed Scene 2:
The business experiences noticeable variations in activity between day and night.
Example:
Business peaks occur every day at 6:30 AM and 5:30 PM, requiring an additional 10 nodes for computing power. The peak duration is uncertain, while a lower computing power requires only 1 node during other times. In this case, you can configure three scaling rules to ensure adequate computing power while optimizing costs.
Rule 1: Set a time-based scale-out rule by choosing to repeat the rule daily and schedule it to be triggered at 6:15 AM tomorrow to scale out 10 nodes.
Rule 2: Set a time-based scale-out rule by choosing to repeat the rule daily and schedule it to be triggered at 5:15 PM tomorrow to scale out 10 nodes.
Rule 3: Set a load-based scale-in rule, and select the monitoring metrics as needed. It is recommended not to set a validity period, with the default being active all day, scaling in 9 node.
Note:
1. The basic configuration sets the maximum number of nodes to 10 and the minimum to 1 to prevent resource wastage and ensure sufficient computing power without any elastic resource shortages. The priority order of the three scale-out rules is Rule 1 > Rule 2 > Rule 3.
2. Scaling out resources takes time, and the time required is proportional to the number of resources being scaled out. It is recommended to prepare resources at least 15 minutes in advance. Typically, the time required is relatively short.