HBase application scenarios

1. User portraits

For example, the user click behavior, browsing behavior, etc. generated by large video websites and e-commerce platforms are stored in HBase to provide data support for subsequent intelligent recommendations.

2. Message/order storage

This scenario is mainly used in e-commerce platforms, because HBase provides a low-latency, high-concurrency access capability

3. Object storage

The object storage here is actually a medium object storage, which is an over-buffering of HDFS storage files, because if we store a large number of small files of 1M or 2M directly on HDFS, it will It puts pressure on metadata maintenance on NAMENODE, so it can be easily over-merged in HBase and then persisted to HDFS. HBase provides medium object storage capabilities, with medium object sizes ranging from 100k to 10M.

4. Time series data

The time series data here refers to data that changes with time, such as speed display, weather, temperature, wind speed, traffic flow, etc.

5. Cube analysis (KyLin)

Use KyLin to build Cubes from data in Hive or kafka. These Cubes will be stored in HBase for other applications or other systems. Real-time query or real-time display.

6. Feeds stream

This scenario is mainly used in Douyin or other small video systems. Feeds stream can be understood as a content aggregator, which can help users Get the latest feed content in real time.