Companies build their own customer data platform to collect and and manage user data from various sources. With this data, they create 360-degree user profiles, analyze user behaviors, interests, and demands as a reference for their marketing and business operations, in order to improve user experience and boost growth.
The above shows the typical workflow of user profiling and behavior analysis. Common solutions are often challenged by the following problems:
  • Inflexible table schema

    As part of production iteration, new fields need to be added to data tables. Solutions that do not support flexible schema change will not be able to serve the needs of the evolving business.
  • Complicated computation

    In the analysis of discrete events, general-purpose SQLs often add to the complexity of computation and introduce large amounts of join operations. Those cause low efficiency in computation.
  • No support for real-time update

    Affected by computing resources and business needs, the hundreds of user tags are generated at various paces, making it hard to generate the "latest" user profiles.
  • Slow query response

    As each user produces multiple events and behaviors per day, the total data size grows exponentially as user base expands. Common solutions often find it hard to provide real-time query response with such huge data.

The VeloDB solution

VeloDB-based customer data platform
VeloDB-based customer data platform
  • Data integration

    • ·VeloDB can extract data from event tracking and databases such as MySQL and Oracle. It supports Flink CDC and high-concurrency small-batch imports to ensure low data latency. It also provides Exactly-Once guarantee in data ingestion.
    • ·VeloDB provides data lakehousing capabilities and allows access to multiple external data sources including Hive, JDBC, and Elasticsearch.
  • Data storage

    • ·To write data from multiple data sources to message queues, VeloDB supports lightweight data cleaning and data modeling in the real-time computation layer via ETL or ELT mode.
    • ·VeloDB Cloud utilizes low-cost and highly available object storage. VeloDB Enterprise supports tiered storage and moving cold data to object storage.
  • Data service & application

    • ·VeloDB supports MySQL protocol and ANSI SQL so it can be easily integrated with data services and applications.
    • ·After user tagging and segmentation, VeloDB can provide interfaces for recommendation and queries to facilitate user profile analysis and precision marketing.

Schema change within milliseconds

Enable real-time column update of flat tables to support frequent changes in user behavior attributes.

Extensive function support

VeloDB provides users with a rich collection of behavior analytic functions that are simple and efficient. It can deliver a query latency of less than 10 seconds in behavior analysis of 300 billion rows of active data.

User segmentation within seconds

Powered by orthogonal bitmap

Try VeloDB now

VeloDB Cloud
Fully managed, cloud-native, real-time data warehouse service
Start free
VeloDB Enterprise
Self-managed software on premises, on VMs, or K8s
Start free