AISWare AntDB helps a Province Telecom big data analysis platform achieve second-level response

Customer requirement

A provincial telecom big data analysis platform needs to conduct rapid statistical analysis on tens of terabytes of BSS data, such as customers/users/accounts, orders, instances, etc. the data level of each analysis is billion-level. Meanwhile, it needs to open this real-time data analysis and query ability to other vendors. The original big data architecture of Hadoop + Hive platform can not meet the needs of customers in terms of timeliness and performance. Customers consider using the open-source Postgres XL database cluster to support real-time query and analysis scenarios, but there are bottlenecks in performance and expansion in the implementation.

Construction plan

AISWare AntDB, which can support 100 TB level data scale of high concurrent transaction business and equal volume level data scale of analytical business, replaces the open-source Postgres XL to build a real-time query platform for big data analysis. Using the shared nothing architecture, several x86 servers with 4C8D1GTM network architecture have completed the deployment of AntDB. The deployment architecture is as follows:

Application effect

The new big data analysis platform uses AntDB distributed SQL optimizer and parallel execution engine to realize the optimization strategy for associating dozens of tables with nearly a hundred fields of group by operating parallel processes between nodes and within nodes. The performance has increased up to a hundred times, and meets customer requirements for analyzing massive data in seconds. Since its launch, it has been running steadily and has solved all the performance and technical problems encountered in the early stage, which has been highly recognized by customers.