Current situation analysis

Bank B’s e-banking self-service analysis system (“the system” for short) needs to query the customer transaction details on the specified date, and the business personnel can set the filtering condition on the interface at will to filter and query in tens of millions of data per day. The query results are used to find the target customer, formulate the marketing plan and evaluate the effect of campaign and so on. However, the existing system can only support five concurrent accesses, which is far from meeting the access needs of a large number of business personnel.

Why does this happen? The reason is that the backend of the system is unable to support more concurrency accesses. The system is directly connected to its backend, i.e., bank B’s central data warehouse. The data warehouse, however, is not specially for the system but shared by the whole bank, and also undertakes other application tasks such as batch running. Since the data warehouse has 48 cluster nodes, almost reaching its upper limit, and cannot be expanded any more, it can only provide the system with five connections, and it is impossible to provide more. Moreover, despite there are five connections, it is difficult to ensure smooth use. Once the data warehouse performs other important tasks, the operation of the system will be very laggy. continue reading →