LindormDFS (LDFS) is a distributed file storage service for large amounts of unstructured data. LDFS shares the underlying storage with multiple Lindorm engines and provides database-level security and reliability. LDFS is fully compatible with open source Hadoop Distributed File System (HDFS) protocols. Therefore, you can use open source HDFS clients to directly access LDFS. This way, LDFS can be seamlessly integrated to the open source ecosystems of big data and cloud computing.
Select storage specifications
If you use LDFS as the underlying storage of data lakes and big data applications, we recommend that you select Capacity for the Storage Type parameter.
You are billed for the storage of Lindorm based on the reserved storage capacity. Make sure that the reserved storage of LDFS exceeds your required maximum storage capacity by more than 10%.
The Capacity storage type is cost-effective and offers the sequential read and write capabilities that provide high throughput. However, the Capacity storage type cannot support a large number of random read requests. Therefore, it is suitable for data storage in big data computing scenarios.
Select node specifications
LDFS nodes handle read and write requests, manage data blocks, and provide support for HDFS-based data communication. You can determine the number of LDFS nodes in your business scenarios from the following two dimensions:
Data size: Each LDFS node can store 10 TB to 50 TB of data.
Bandwidth: Each LDFS node supports a data transmission bandwidth in the range from 100 MB/s to 200 MB/s.
You can individually calculate the required number of LDFS nodes based on the data size and bandwidth in your business. Then, you can choose the larger value from the calculation results as the number of LDFS nodes.