With the rapid development of UFIDA's marketing business, the database used previously encountered some pain points and appeals, such as complex business queries that were slow and some core functions that timed out during queries; At the end of the month, quarter and year, business queries involve a large number of queries and high concurrency, which requires a rapid jump in database resources and a jump in promotion; The database reading delay is too large to be fully utilized and the expansion ability is insufficient; The migration database requires high compatibility and capacity.

In view of the above pain points, UFIDA finally chose Alibaba Cloud's PolarDB database after full research on mainstream databases. Test certificate, PolarDB parallel query can effectively improve the execution speed of complex queries. Under the marketing cloud verification scenario, the average query time is reduced by 66%, and the problem of query timeout is solved. PolarDB has the flexibility of minute level. It can increase nodes in 5-10 minutes, and vertically upgrade nodes in 15-20 minutes, which can effectively meet the demands of marketing cloud elastic upgrade. At the same time, The PolarDB consistency level setting solves the problem of database read latency and makes full use of the database read.

In the face of compatibility and performance challenges that may arise from cross platform and cross version upgrades, Alibaba Cloud database autonomic service DAS is used in the scheme. The compatibility and performance are evaluated through DAS intelligent pressure testing to ensure that there is no compatibility and performance risk after upgrading from the original database to PolarDB. The scheme collects a full amount of SQL from the production environment, processes and learns the message flow and traffic, and finally falls into OSS to form a compressed file, which can be customized for playback and pressure testing. This solves the cross version compatibility problem of databases and the efficiency problem after migrating databases. Although it seems that the process is very long, you only need to configure the entire migration process on the page. Finally, it can capture real business traffic under low load, write traffic playback pressure test, automatically generate pressure test traffic, and automatically convert syntax.

UFIDA's financial cloud is a multi tenant form. Some tenants' data and access are skewed seriously. There is a single database and single tenant scenario, and the horizontal splitting effect is not good. And there are many single table columns, SQL is complex and heavily dependent on databases. At the end of the month and the end of the quarter, the amount of data accessed is large and the concurrency is high. The transformation from the traditional database requires that the database product part has the capability of commercial database, and has high requirements for the stability of the product and the continuous investment of the product team.

After many in-depth exchanges with Alibaba Cloud architects, product architects, product research and development, and product managers on site, UFIDA selected PolarDB. Through parallel query and high flexibility of PolarDB, the consistency reading ability can meet the current business needs. And co created column memory, bypass memory Filter Index、 Product capabilities such as multi write, operator push down, and query depth optimization meet the design and planning of financial cloud architecture.

UFIDA's supply chain cloud includes supply chain cloud, new retail, manufacturing cloud and other business fields, with large amount of data and complex business query model. Taking the supply chain cloud as an example, the database instance involves 2000+tenants, 30000+tables, dozens of business types, and more than 10 complex query scenarios, such as sales analysis, retail collection details, business status query, store goal achievement, commodity inventory query, receipt/issue details, sales shipment push list, material reference list, sales shipment list query, sales issue query Sales receipt query, etc. There are high requirements for the storage capacity, query performance and stability of the database.

After several technical discussions with Alibaba Cloud, it was confirmed that the enterprise level capability IMCI (inventory index) built into PolarDB could solve the demand of supply chain cloud for improving query performance. Through continuous verification, the listed index has been launched, achieving the following effects: the performance of complex query of 5-10 tables with 100 million level data has been improved by 4-40 times, the comprehensive cost has been reduced by 40%, the performance of monthly closing and inventory counting has been improved by 40%, and the slow query has been reduced by 30%.

The iuap cloud platform supports the business operation of the Yonyou cloud service. In the daily operation of the business, various slow queries and high concurrent queries will occur, which will lead to performance exceptions or potential performance problems in the business and infrastructure database. The traditional way requires manual processing by the DBA, which is costly and difficult. After using Alibaba Cloud DAS, we can help DBA deal with abnormal SQL more efficiently and improve the operation and maintenance efficiency of the DBA team by 30% through the functions of the DAS system such as SQL flow restriction, automatic SQL optimization, and solidified execution plan.