As a leading pharmaceutical distribution enterprise in China, Jointown adheres to the concept of "Technology makes service more excellent" and has made many attempts to improve its technology and service, of which, FBBC business is the representative of Jointown's innovation in the pharmaceutical + Internet health industry chain, where "F" stands for upstream pharmaceutical companies, the first "B" stands for Jointown, the second "B" stands for terminal pharmacies and clinics, and "C" stands for consumers.

According to Wang XiaoHai, technical director of Jointown’s e-commerce business platform, FBBC aims to set up an F2B2B2C platform connecting pharmaceutical companies/suppliers, end customers and consumers to realize rapid distribution of products to a wider range of terminals.

Through B2B pharmaceutical platform and "Medical Circulation" APP platform, Jointown effectively integrates and manages pharmaceutical resources to realize a perfect system benefiting pharmaceutical production enterprises, Jointown, terminal pharmacies/clinics, and consumers, thus providing highly cost-effective services for the pharmaceutical and health industry. At present, more than 7 thousand pharmacies and more than 70 thousand sale-people have been conducting business through Jointown’s "Medical Circulation" APP platform to inquire about drug prices and stocks and other information at any time to support business decisions.

In the meantime, the "Health 998" of Jointwon is a third-party B2C platform providing online sales tools free of charge for pharmacies. While helping pharmacies to obtain higher economic benefits, it also attracts them to purchase more through the Jointwon’s platform. "Mendiantong" is the backend dedicated to pharmacies where 8,000 pharmacies will settle in 2019, covering provincial capitals and core cities around the country. "yyjzt.com" is the backstage supporting the online B2B business as a whole, providing search, order placing, online payment, backend operation management and other functions.

Currently, all of these businesses are running on Amazon Web Services, including "Medical Circulation", "Health 998", "Mendiantong", and "yyjzt.com". "We finally chose Amazon Web Services Cloud after various comparative tests and comprehensive consideration from the perspectives of ease of operation and maintenance and innovation of tools," said Wang Xiaohai. "Amazon Web Services’s excellent technology as well as rigorous and professional support services are impressive. "

Jointown adopts the Amazon Web Services cloud services including Amazon Elastic Compute Cloud (Amazon EC2), Amazon Simple Storage Service (S3), Amazon Virtual Private Cloud (Amazon VPC), Amazon Aurora, Amazon Lambda, Amazon Elastic MapReduce (Amazon EMR), Amazon ElastiCache, Application Load Balancer (ALB), Amazon CloudWatch, etc.

Among the Amazon Web Services services used, Jointown was impressed with Aurora. The businesses of Jointown’s B2B system are characterized by reading more and writing less, with a ratio of reading to writing between 8:2 and 7:3, and a certain degree of volatility. Influenced by promotional activities, working hours and other factors, a large gap between wave peaks and wave troughs often appears (as shown in Fig. 1).

Jointown can easily realize the separation of reading and writing of the database and the expansion on demand with Aurora. Using Aurora Database Service, a single Aurora cluster can support up to 15 read-only nodes and support online auto-scaling and auto-shrinking. Jointown routinely uses 2 Aurora database nodes, one master and one slave, to cope with regular access traffic, as shown in the blue horizontal line in Fig. 1, and sets a horizontal scaling strategy for the slave node (read-only node). When the server’s CPU utilization rate or number of connections and other indicators are greater than the set monitoring threshold, Aurora will gradually add one or more read-only nodes to meet the needs of peak access traffic in a timely manner. After the peak period, the system can automatically shrink the number of database nodes to 2 according to the set policy, thus reducing the occupation of resources and achieving cost savings. In case of a self-built MySQL database scenario, because of its inflexibility of scaling out, DBA (administrator of database) needs to configure a sufficient amount of database server resources in advance to cope with possible peak access (as shown by the red horizontal line in Fig. 1); some server resources are idle after a peak of access state, thus wasting resources and increasing costs.

The data replication delay between the master library and slave library of self-built MySQL often exceeds 1 second, thus all read operations can not be placed in the slave library, only some report type queries, and queries insensitive to time requirements can be placed in the slave library for execution, resulting in high pressure on the master library and poor separation of read and write. After using Aurora, the data latency between master and slave nodes basically stays about 20 milliseconds, thus more query operations can be executed in the slave library, and effectively reducing the workload of the master library.

The durability, availability, security, usability and other aspects of databases are also of great concern to Jointown. Aurora's features like replicating 6 copies of data across 3 Availability Zones, deploying database instances across Availability Zones, automatic fail over, automatic backup, and point-in-time recovery are well received by the technical team. After using Aurora, the database performance of Jointown B2B has improved as a whole by 5 times and TCO has been reduced by 50%.

With ALB, Jointown has realized grayscale release of business system version updates. When releasing, start an instance of the production environment, switch over the new code and traffic, and then stop the old version of the virtual machine to complete the release of a new version. Wang XiaoHai said that in the past, every time a new version was released, the team had to prepare from 8 p.m. until around midnight to avoid the peak. Now, grayscale release can be done at any time and there is no need to work overtime for it. Jointown's numerous systems are released as many as 100 times in a month, large and small. Through grayscale release, the time and energy of team members are greatly saved.

Serverless computing is the development trend in the future. Through the Lambda serverless service, Jointown has deeply experienced the construction method for a new generation of cloud-based application. Jointown has used Lambda to build an application to automatically runs scripts for checking middleware and applications, find out the codes that call the interface frequently, verify and guarantee system stability. The Lambda serverless service only requires to pay for each check, which not only makes it unnecessary to spend energy managing one additional server, but also saves the expenditure for server idle time. Wang Xiaohai said that Amazon Lambda is easy to use and very user-friendly.

Jointown mainly uses the Amazon Web Services China (Ningxia) Region operated by NWCD, with its Amazon Web Services-based system architecture as shown in Figure 2.

Wang Xiaohai concluded that with the Amazon Web Services Cloud, Jointown has achieved the following benefits:

First, the cost saving has been achieved. By using Aurora Database, Jointown’s B2B database performance has been improved by 5 times as a whole, TCO has been reduced by 50%, achieving cross-AZ deployment, load balancing/automatic fail over, fine monitoring/on-demand auto-scaling, and other functions.

Second, efficiency has been improved. With the release of new versions as an example, more than 100 new versions are released every month, which can be carried out at any time, and thus there is no need to work overtime for 5 or 6 hours each time anymore. The reduction in overtime hours of more than 11 man-day workload per month has greatly increased productivity.

Third, Amazon Web Services is reliable and stable, providing a solid guarantee for the operation of business operation.

At the same time, Amazon Web Services’s support services are very timely and in place. Jointown is able to continuously deepen the application of new technologies and services and continuously improve the application level in terms of prototyping, daily follow-up, architecture reconstruction and optimization.

Wang XiaoHai said that Amazon Web Services can provide strong support for tools and infrastructure so that data warehouse and deep learning framework are be the next application directions. In the future, based on Amazon Web Services, Jointown will further expand the applications of big data and AI, such as customer portraits, demand forecasting, smart recommendations, etc.

Jointown, a pharmaceutical distribution company with annual sales revenue of CNY 100 billion, is to be commended for its pioneering efforts in the wave of cloud computing. As technology makes service excellent, we hope that the superiority of cloud computing will enable Jointown to innovate like a start-up company and establish a unique competitive advantage.