In the age of mobile Internet and social media, users record their words and behavior on the Internet anytime and anywhere. How enterprises use such data is an essential part of business. User personas are built based on these words and behavior. User persona technologies have been widely used in business scenarios such as precision marketing, system recommendation, advertising, risk control, and intelligent customer service. User persona data has the following characteristics: Large amounts of data is generated. The data is highly concurrently read and written. Data needs to be archived. The analysis results for large amounts of data need to be returned. The data is collected from dynamic columns. User persona data must support multi-dimensional and complex queries.
ApsaraDB for Lindorm (Lindorm) provides a storage system for large amounts of semi-structured and structured data. It is perfectly suited for the characteristics of the user persona workloads for the following reasons: Transactions are not required. Large amounts of data and highly concurrent reads and writes need to be handled.
The following figure shows the architecture.
1. Lindorm is cost-effective. It provides the hot and cold separation capability that allows you to separately store hot and cold data from the same table. Lindorm provides optimized compression algorithms to optimize the data compression. Lindorm also allows you to store multiple types of data.
2. Lindorm provides the high throughput capability. In various scenarios, Lindorm provides a throughput performance that is multiple times higher than the performance provided by Apache HBase Community Edition 2.0.
3. Lindorm allows you to archive incremental data in real time. You can use Lindorm Tunnel Service (LTS) to archive data to offline storage media in real time.
4. Lindorm provides the BulkLoad feature. Lindorm uses the log-structured merge (LSM) technical architecture that allows you to efficiently load data. This has little impact on the online business.
5. Data is collected from dynamic columns. The user persona data structure changes frequently. Lindorm allows you to use dynamic columns to collect user persona data.
6. Lindorm supports multi-dimensional and complex queries. Lindorm uses a cloud native architecture that supports global secondary index-based queries. The Lindorm search engine allows you to run multi-dimensional queries.
Solution details and expert support
For more information, see A Lindorm-based solution for using big data to build user personas.
If you have questions, you can ask an expert. For more information, see Expert support.