Recently, Tencent Cloud officially released a new object storage product- intelligent tiered storage , which can automatically convert the hot and cold tiers of data according to the access mode of user data, providing users with a low-latency and high-throughput product experience consistent with standard storage. At the same time, it has lower storage costs.
Users familiar with data storage know that data with high access frequency is hot data, and data with low access frequency is cold data. Hot data uses standard storage types to store, which can get a better access experience; while cold data generally uses low-frequency storage or archive storage types, which can save storage costs.
With the development of business, hot data will cool down, and cold data may also be warmed up again. Although the current object storage COS supports the specified rules for object configuration to realize the hot and cold conversion of data, the sinking of data through the life cycle is one-way Yes, it is only allowed from standard storage> low frequency storage> archive storage> deep archive storage. It also supports step-by-step settlement (such as standard storage> archive storage). Reverse direction is not supported. After the data is transferred to low frequency storage, you need to pay more for access Data retrieval costs, and the life cycle can only be settled for buckets and designated prefixes at present. Only the life cycle settlement data can not deal with scenarios with irregular access patterns.
Intelligent tiered storage application scenarios
Compared with the standard storage configuration life cycle method, intelligent tiered storage provides a hot and cold tiering mechanism for data, which is mainly used to automatically convert data to hot data when the access frequency increases, and automatically switch back to cold data when the access frequency decreases. , Thereby reducing the storage cost of user data, intelligent tiered storage is suitable for data whose access patterns are not fixed or whose access patterns cannot be estimated.
In the application scenarios of most industries, data generally has the characteristics of staged hotspot visits. After a period of time, more than 80% of the data visits decrease. For example, video tutorials, product pictures, activity files, electronic courseware, video resources and other data in industries such as film and television media, e-commerce, online education, games, cultural tourism, etc. The number of visitors was a lot when they first went online, but gradually fewer people visited , But the resource is still online and will still be accessed. If you choose standard storage, the storage cost will be high; if you choose low-frequency storage, you will not only need data retrieval costs, but also damage the user's access experience. However, if you choose intelligent tiered storage, you don’t need to pay attention to the popularity of data access, and the object storage service monitors the frequency of data access, and the data can be intelligently converted between hot and cold access tiers, without data retrieval costs, without harm While the user accesses the experience, it effectively reduces storage costs.
Intelligent tiered storage advantages
1. No need to pay attention to data access mode, intelligently transform storage layer
You only need to specify the object storage type for the data to apply the intelligent tiered storage feature. When the user uploads data, choose to store it in COS in the smart tiered storage type. The data will be temporarily stored in the frequent access layer, and then COS will periodically monitor the number of data accesses, and transfer the data to storage when there is no data access for a period of time Lower cost access layer. If the data is accessed again, it will be transferred to the frequent access layer again to ensure data read performance.
2. There is no need to retrieve low-frequency data, it can change to high-frequency data every second
Intelligent tiered storage is a storage type at the file object level. COS will intelligently monitor each object in that storage type. When it detects that the access of an object decreases, it will automatically convert the data to low-frequency storage. After being accessed again, COS will directly convert the data to standard storage without having to pay any data retrieval fees.
3. Data is automatically settled, and the storage cost is better
By monitoring the frequency of data access and automatically managing the data access layer, the intelligent tiered storage type also participates in the object storage life cycle process, and users can settle the smart tiered storage into archive storage on demand, further reducing the cost of data storage in the cloud.
4. Stored data is stable and durable, and supports disaster recovery in the same city
Intelligent tiered storage uses erasure code redundant storage, providing data reliability of up to 99.999999999% (11 9s); data is stored in blocks, and concurrently read and write, providing up to 99.95% business availability. The same-city disaster recovery MAZ architecture has synchronized intelligent tiered storage types, with data design reliability up to 12 nines, and service design availability up to 99.995%.
Through data hot and cold tiered storage, intelligent tiering can help users find a balance between storage cost and read and write performance . Intelligent tiered storage, as a storage type, naturally adapts to COS's API, SDK, tools, and ecological applications, facilitating users to manage data stored in the cloud on demand. At present, COS has provided external storage products covering different business needs such as standard storage, low-frequency storage, and intelligent tiered storage. In the future, it will continue to iterate and introduce more products and functions to provide users with the most cost-effective storage services in the industry.