Building an Analyst Platform for a Smart, Open, Digital ICBC
Challenges
-
Low Performance
A query takes hours to complete on average.
Real-time data processing capabilities are weak.
A query takes hours to complete on average.
Real-time data processing capabilities are weak.
-
Difficult O&M
A fault in the old platform can have global impact.
The old platform is unstable and difficult to manage.
A fault in the old platform can have global impact.
The old platform is unstable and difficult to manage.
-
Limited Capacity
The old platform is not so scalable.
Its computing power and capacity are limited.
The old platform is not so scalable.
Its computing power and capacity are limited.
Solution
Core data warehouses and data marts are constructed based on GaussDB(DWS), providing enterprise-grade multi-tenancy and resource isolation capabilities to support batch processing and hybrid load of online services.
An independent analyst platform is deployed. It has 480 nodes in a single cluster, responding to the analysis requests from more than 13,000 analysts in real time. R&D efficiency is significantly improved.
A converged data analysis solution is developed based on GaussDB(DWS) and FusionInsight MRS. Structured and unstructured data is integrated to facilitate management. Real-time, quasi-real-time, minute-level, and hour-level computing are supported.
AI and knowledge graphs enables the data platform to evolve to a big data & AI platform.
Customer Benefits
-
High Query Efficiency
The average query waiting time was reduced from 300 minutes to 1.5 minutes.
The average query waiting time was reduced from 300 minutes to 1.5 minutes.
-
Ease of Use
The new platform is centralized. The number of devices that can be managed per person increased tenfold.
The new platform is centralized. The number of devices that can be managed per person increased tenfold.
-
No Capacity Bottlenecks
Online operations of 13,000+ analysts
Analysis on 10 PB data
Online operations of 13,000+ analysts
Analysis on 10 PB data