Sensors, crawlers, lidar, cameras and other front-end equipment and software, as well as a large number of users, are inputting a large amount of unstructured data into enterprises every day. In order to preserve and maintain data, a new factor of production, enterprises pay a rapid increase in the cost of unstructured data storage every year.

For most enterprise users, the data is periodically accessed. After a certain period of time, more than 80% of the data gradually cools down. The access performance of hot data is high. After a certain period of time, the hot data becomes cold and the application accesses the cold data at a low frequency.

How to solve the massive unstructured data storage and access performance, while taking into account the overall use cost of unstructured data by enterprise users, is the main problem facing CIOs.

YRCouldFile file storage system of intelligent hierarchical function, can according to user needs, the custom of cold and hot data strategy, data automatic cold flow to low cost object of public cloud storage and completes the compression, still up for business to provide standard file access interface, and maintain a directory structure remains the same, the data flow between hot and cold data layer completely transparent to the business, Effective balance between cost and performance. Recently, Yan Rong Technology released a new version of YRCloudFile, which comprehensively upgraded the intelligent layering function and refined the layering policy to the directory level. For example, cold data in directory A is flushed to Ali Cloud OSS, cold data in directory B is flushed to AWS S3, and cold data in directory C is flushed to local object storage.

What are the benefits of a fine-grained layering strategy?

Cold data definitions are more flexible

As we all know, the object storage cost of public cloud consists of storage capacity cost, API call cost, and network inflow and outflow cost. In many cases, the cost of API calls or network inflow and outflow is much larger than the cost of storage capacity.

Data cooling is a gradual process, in the process of data cooling, may still be accessed, this part of the cost also adds up to an expense. YRCloudFile directory level intelligent hierarchical function launched, can be a good solution to this problem. It helps users distinguish hot and cold data. For cold data, it may be accessed only a small amount when stored in the local object storage. For colder data, you can set up a directory to connect to an unlimited public cloud object store for archiving, and this part of the data is basically no longer invoked.

So easy, different applications, different strategies, different object storage vendors

For data centers, different applications have different definitions of cold data and different requirements for data storage. For example, cold data with high data security requirements must be stored locally. Data with low data security requirements can be stored in the public cloud. For example, training data will no longer be accessed after 2 weeks of frequent training, and will become cold. Training result data will need to be accessed frequently for a long time.

We also need more flexible and simpler policies based on different application types and data security considerations. YRCloudFile defines different cold data policies and cold data storage locations based on different scenarios and requirements.

Capricious, save where you want to save

Cold data is no longer bound by any vendor. You can store cold data on major public cloud object storage platforms around the world, private commercial object storage, or open source Ceph. It is not difficult to interconnect YRCloudFile with different object storage platforms. You can use a simple UI to interconnect YRCloudFile with different object storage platforms.

YRCloudFile is constantly optimizing the management of the life cycle of massive data while pursuing the ultimate performance. In the future, we will implement more innovative features in data lifecycle management, so stay tuned… 😄

Object storage support video tutorial, please stamp: www.bilibili.com/video/BV16b…