Do you have more members? ——The incremental innovation of database helps to promote the practice of business brain at multiple points

Di... "Do you have more members?"

For people who often go to large supermarket chains such as Wumart and Metro, the Di Di sound of scanning the code and the following sentence should be very familiar. But as a professional supplier of digital systems for supermarkets, what Duodian does is not just as simple as a checkout. Under the support of the new business-finance integration strategy, the multi-point Dmall OS is not only a system that supermarket customers can use every day, but also a system that CFOs and CEOs pay attention to every day.

The wonderful picture brought by the integration of business and wealth to the enterprise
Multipoint is a digital solution provider for new retail. Its Dmall OS product is a full-scenario cloud solution that integrates people, goods, and markets, and it is also a flagship product of Multipoint. Based on this, the next step is to provide retail enterprises with a business brain with the ability to integrate business and finance.

In the past, enterprises needed to realize business digitization through a large number of business applications based on scenarios and processes, but at the same time, the financial data that the enterprise management was more concerned about were generated by financial systems with completely different operating logic, statistical caliber, and statistical methods. This disconnect between business and finance also makes it difficult for companies to quickly grasp current operating data and adjust business strategies and operating policies through timely financial feedback. The so-called integration of business and finance is to break down the barriers between the two systems, so that changes at the business level can be directly reflected in the real-time financial statistics, so that enterprises can gain flexibility and agility in the fierce market competition.

Judging from the effect after implementation, the integration of industry and finance is wonderful enough for enterprises, but in the actual system construction process, the realization of the integration of industry and finance is full of challenges.

Taking the retailer super-industry targeted by Dmall OS as an example, its business side corresponds to a massive number of transactions and a huge and scattered number of stores. To provide support for basic businesses such as commodity management, cashier, and membership, Dmall OS needs to be equipped with a powerful OLTP database. The various analysis functions required by the financial side are typical OLAP applications. Therefore, before the business logic is straightened and the system connection is completed, a large amount of data needs to be imported from OLTP to OLAP. The key real-time requirements in the concept of integration of business and finance mean that Dmall OS must ensure the performance and reliability of the OLTP database on the one hand, and complete the real-time data import, real-time synchronization, and real-time analysis on the other hand. The difficulty of implementation can be imagined. .

After understanding this difficulty, we can easily understand why the integration of business and wealth in many enterprises cannot be real-time, but can only be asynchronous.

However, Dmall OS has crossed these technical thresholds and has been recognized and favored by a series of top users in the industry such as Wumart and Metro. At the bottom layer, the enabling tool that helps Dmall OS realize the key transformation of the integration of business and finance is TiDB.

PingCAP's TiDB, multi-point integration of business and wealth
In fact, database challenges encountered by multipoint are not uncommon.

On the one hand, the basic businesses represented by cash registers and inventory correspond to typical OLTP database applications, and the huge sales volume of the supermarket format makes this part of the business have high requirements for performance and stability. When meeting this part of the business needs, like most Internet companies, Duodian chose the open source MySQL at the beginning. Its good performance, rich ecology, abundant talents and convenient secondary development are its biggest advantages. However, as a technology born in the 1990s, MySQL still cannot achieve a good balance between "performance degradation caused by data volume growth" and "ensure performance through complex and high-risk database and table operations".

On the other hand, in order to realize the integration of business and finance, the multi-point Dmall OS also needs a high-performance OLAP database that can provide support for report consolidation and massive data analysis. And, in order to maintain the overall open source of the software stack and the operational coherence of business people, the new database also needs to be open source, and preferably has a similar operational logic and syntax to MySQL.

Of course, if the multipoint just uses another set of OLAP database to meet the financial analysis needs and bear the increased operation and maintenance cost brought by the dual database, then the story is over. But what TiDB provides for multipoint is a completely different path.

As a database with HTAP capability, TiDB can meet the needs of two different applications, OLTP and OLAP. When faced with the OLAP requirements of multi-point business-finance integration, TiDB can provide high-performance analysis capabilities to meet the financial-side report consolidation and analysis needs of business-finance integration. With the powerful TiFlash columnar storage engine, TiDB can get results in about 5 seconds when facing a large table with 680 million rows and a full table aggregation query, and it only takes about 38 seconds for a full table aggregation of 4 billion rows in a large table. of OLAP business has also reached the real-time level.

Business and financial integration structure diagram

The HTAP capability of TiDB means that multiple points can first deploy TiDB in the OLAP field to solve existing pain points. After you have accumulated enough operational and business experience, you can switch to TiDB at a lower cost and risk at multiple points, realize the unified architecture at the database level, simplify operation and maintenance, and lay a solid foundation for further data platform construction.

As a new-generation cloud-native database, TiDB not only has extremely high execution efficiency, but also supports users to easily cope with the growth of data volume and business needs through clustering and horizontal expansion, avoiding the huge workload and risks caused by traditional database sub-database and sub-table. . As an open-source product, TiDB not only conforms to the overall technical route of Multipoint, but its active community and strong original factory support also allow Multipoint to face future changes without being bound. At the same time, in terms of operation and syntax, TiDB also tries to be consistent with MySQL, so that the user's operating experience and usage habits can be unified.

In actual deployment, high-speed data synchronization is achieved between MySQL, which undertakes OLTP business, and TiDB, which undertakes OLAP, through the TiDB DM tool and related APIs developed by PingCAP, and financial-level data consistency is ensured.

With the powerful functions and performance of TiDB itself and the all-round support of the original engineers, Multipoint not only allows the business-finance integration service to grow and expand together with customers through the horizontal expansion of the cluster, but also completely solves the problem of the business-financial integration service. Corresponding HTAP requirements. In daily operation and maintenance, the expansion capability of TiDB also frees the multi-point operation and maintenance personnel from the tedious operation of sub-database and sub-table, which greatly reduces the cost and workload of multi-point database operation and maintenance.

When faced with new technologies and new products, companies' choices are usually conservative, because this is related to business stability, especially at the database level. However, the deployment method of using a new database for new services and mature products for old services at Multipoint proves that TiDB's HTAP capabilities, diverse functions, and strong adaptability can provide enterprises with a steady stream of incremental innovation. The path allows enterprises to have more room for transformation and adaptation in the key link of digital transformation, "database switching".

A good system should "love the new and never tire of the old"
Starting from the simplest cash register system, to the reconstruction of people, goods, and fields in the digital space, to the more advanced integration of business and wealth and the management brain, the multi-point development path along the way is very clear: do a good job in the present At the same time, each time a little more, a little more. On the premise of stability, through gradual functional and structural evolution, until the complete construction of various functions of the operating brain. The development logic of multi-point is not only visible and palpable, but also able to convince the major supermarkets and enterprises. Therefore, we can see that a number of industry giants are willing to use the multi-point Dmall OS as their own business brain, and deliver business, finance, and decision-making with confidence.

Multi-point incremental innovation also proves that PingCAP's TiDB not only has advanced performance and architecture, but also has complete tools, interfaces and service support, which can be calm in joint operations with traditional databases. For the majority of users, such an application method provides an excellent opportunity for low-cost, low-risk and rapid access to new technologies.

It can be salty or sweet, and the new is never tired of the old. This is the secret of long-lasting business operation, and it is what the new generation of data technology represented by TiDB should look like.

"New Economy DTC User Scenario White Paper" explains the data challenges and data architecture design in five scenarios of the new economy, and analyzes the best practices of leading enterprises such as new catering, new retail, commercial real estate, high-tech manufacturing, and tourism hotels