Cost optimization

Convert storage classes based on data access frequency and response time

When you identify decreased access frequency to your dataset or reduced requirements for immediate data retrieval, you can set reasonable lifecycle rules for the data to automatically transfer it to a more cost-effective storage class without affecting data access performance.

You can configure lifecycle rules for specific data based on different data characteristics and usage frequencies by filtering objects using prefixes or tags. For cold data that is rarely accessed, you can automatically migrate it to more cost-effective Archive Storage. For hot data that is frequently accessed, you can keep it in the Standard storage class for faster reading speed. This not only helps optimize storage costs but also improves data access efficiency.

Example

User A stores all 100TB of business data (10 million objects) in a directory called "dir" in a bucket in China (Hangzhou) using the Standard ZRS storage class, mainly for the company's daily operations. In the early stages of the business, data access is very frequent. However, as the business continues to evolve, the data access patterns also change accordingly. To optimize costs, User A wants to adjust the storage class of the business data according to different business stages.

Frequent overall data access in the early stages of business

In the early stages of the business, due to frequent data access, the 100TB of data is accessed an average of 5 times per month. At this time, the user's need for data access speed is critical, so it is not advisable to choose Archive Storage, which cannot be read in real time. User A is considering continuing to use Standard storage or configuring lifecycle management rules to convert some data to Infrequent Access.

• The cost of Standard storage is 0.15 CNY/GB/month, GET requests within 20 million are free, and those exceeding this limit are charged at 0.01 CNY/10,000 requests, with no data retrieval fees.

• The cost of Infrequent Access is 0.10 CNY/GB/month, GET request fees are 0.1 CNY/10,000 requests per month per region, and data retrieval fees are 0.0325 CNY/GB.

Cost analysis:

From the results in the table above, it can be seen that when data is accessed frequently, Standard storage has advantages in both cost and access speed, ensuring efficient data access while avoiding additional reading fees in the Infrequent Access mode. Therefore, using Standard storage is the most suitable choice for business data at this stage.

Decreased access frequency for some data after business growth

As the business expands, although the total data storage remains at 100TB, the data access pattern changes with business development: 90% of the data (90TB) is accessed only once a month, while the remaining 10% (10TB) still maintains a high access frequency. To address this change and control costs, User A considers two strategies:

1. Continue using Standard storage: Maintain the status quo and continue using the Standard storage class.

2. Apply lifecycle rules: Convert data with reduced access frequency to the more economical Infrequent Access type to reduce storage costs.

Cost analysis:

1. We can calculate the cost of using Standard storage for all business data as shown in the following table:

   Storage class	Total data	Storage fee (CNY/month)	API request fee (CNY/month)	Data retrieval capacity fee (CNY/month)	Total fee (CNY/month)
   Standard ZRS	100 TB	15360	30	0	15390

2. Use lifecycle rules to filter business data that is accessed an average of once per month through prefixes or tags, amounting to 90TB. Configure this part of the data to automatically convert its storage class from Standard to Infrequent Access within 30 days after the last access. The remaining 10TB of business data still uses Standard storage.

   • Standard storage: Storage fee is 0.15 CNY/GB/month, no API request fees for GET requests within 20 million, and no data retrieval fees.

   • Infrequent Access: Storage fee is 0.10 CNY/GB/month, GET request fees are 0.1 CNY/10,000 requests per month per region, and data retrieval fees are 0.0325 CNY/GB.

   Storage class	Total data	Storage fee (CNY/month)	API request fee (CNY/month)	Data retrieval capacity fee (CNY/month)	Total fee (CNY/month)
   Standard ZRS	10 TB	1536	0	0	1536
   IA zone-redundant storage	90 TB	9216	90	2995.2	12301.2

In the scenario using lifecycle rules, the total cost combining the two storage classes is 13837.2 CNY, saving about 10.1% compared to using Standard storage. Given that the access frequency has decreased, configuring lifecycle rules to convert part of the business data to the Infrequent Access type is an ideal low-cost choice, which can effectively control costs while maintaining data access speed.

Recommended action:

User A can configure lifecycle rules that specify files matching the prefix to automatically convert their storage class from Standard to Infrequent Access within 30 days after the last access, achieving a balance between cost optimization and data access speed. For specific steps, see Lifecycle rules based on the last access time.

Business maturity stage, considering long-term retention of cold data

As the business gradually matures, User A finds that the utilization rate of some business data decreases, with business data being accessed once per quarter on average. User A confirms that the demand for accessing this part of business data in the next few months is extremely low, and the user can accept a brief data restoration time. At this point, User A considers configuring lifecycle rules to convert this part of data from the current Infrequent Access to Archive Storage.

Using lifecycle rules to filter business data that is accessed once per quarter on average through prefixes or tags, amounting to 90TB. Within 30 days after the last access, automatically convert its storage class from Infrequent Access to Archive Storage. The remaining 10TB of business data still uses Standard storage due to frequent access.

• Standard storage: Storage fee is 0.15 CNY/GB/month, no API request fees for GET requests within 20 million, and no data retrieval fees.

• Archive Storage: Storage fee is 0.033 CNY/GB/month, GET request fees are 0.1 CNY/10,000 requests per month per region, and data retrieval capacity fees are 0.06 CNY/GB.

From the table above, it can be seen that if you further configure lifecycle rules to convert this part of business data from Infrequent Access to Archive Storage, the total cost combining the two storage classes is 6450.3 CNY, which is 53.4% lower than the cost of configuring lifecycle rules for Infrequent Access, achieving more significant cost optimization.

Recommended action

If your access pattern is known and fixed, specify all files matching the prefix in the rule, and modify the configuration to automatically convert their storage class from Standard to Infrequent Access within 30 days after the last access, and then automatically convert the data's storage class from Infrequent Access to Archive Storage 60 days after the last access. At this point, the data is automatically converted from Standard to Infrequent Access 30 days after the last access, and then to Archive Storage after another 30 days, achieving optimal storage costs. For specific steps, see Lifecycle rules based on the last access time.