"Hold on! Hold on! Win! The Jingdong preparation team, which was staring at the data screen, was full of joy in an instant, and with the arrival of the last round of red envelope interaction around 0:20 minutes, JD.com's first "Spring Festival Gala Journey" came to a successful end.
Chinese New Year's Eve the JD cloud technicians who participated in the front-line duty on the same day celebrated the successful completion of the task
Compared with the sniper war of online traffic competition in the past, Jingdong can be said to overcome the "difficulty and difficulty" of "traffic + performance", not only calmly coping with the extreme concurrent traffic baptism of up to hundreds of millions of peaks; it also orderly shouldered the overall supply chain performance of retail and logistics in the national "Spring Festival", and properly refreshed the world's most complex technical scene record at the Spring Festival Gala, which is a must!
Looking back at the various moments of preliminary preparation, this year's Spring Festival Gala red envelope interaction has really given JD.com a big problem! "In this process, Jingdong's technical system needs to support red envelope interaction and shopping transaction scenarios, the two scenarios are very different, the peak is frequently switched back and forth, and the technical challenges are great; in addition, in addition to the red envelope interactive scene, it is the New Year Festival, and the Jingdong technology system also needs to support JD.com's transactions, payments, customer service, sorting, distribution and many other business scenarios, with long links. The addition of each link node greatly increases the complexity and difficulty of the project's technical system. This technically challenging 'Everest' is not easy to climb! The members of the preparation team said one after another.
But what is amazing is that in just 19 days, without increasing computing resources and independent support, more than 10,000 technicians based on the JD cloud R & D performance platform to work together, with the cloud native digital infrastructure that has been cultivated for many years and the hybrid multi-cloud operating system cloud ship that has been tempered by many large-scale scene technologies, using more than 70 data centers across the country, successfully building a super-elastic, efficient and agile digital base for the world's largest interaction. Successfully coped with the highest peak of network interactive traffic and the most complex application scenarios in history.
Jingdong cloud technicians who are fighting in the front line
4 hours of continuous interaction, the cumulative number of interactions reached 69.1 billion times, tens of thousands of technical personnel efficient collaboration, 16 seconds of accurate transfer, smooth switching of seconds of resources... What can JD Cloud achieve?
"JD for the first time! Spring Festival Gala red envelope interaction, we have set up a military order! ”
"This is the first time that Jingdong has independently supported such a big project as the Spring Festival Gala Red Envelope Interaction, and it must fight a beautiful battle!" It's easier said than done! According to the technical leader of the Jingdong Red Packet Project Team, "The Spring Festival Gala interactive project is not only a very challenging technical problem, but also a problem of efficient multi-departmental collaboration, and problems such as tight time, urgent tasks, large challenges, and complex personnel must be dealt with." To this end, we have specially pulled through dozens of first-level departments in a short period of time, and more than 10,000 R&D personnel have joined forces to achieve orderly docking and tuning of multiple business systems, so that the entire project can be efficiently coordinated. "At the beginning of the preparation, Jingdong has formed a perfect preparation system structure: unified deployment by the general commander of the preparation, follow-up of the project preparation team, and unified coordination of all research and development teams across retail, science and technology, logistics and other departments.
Engineers of JD Cloud Product R&D Department discuss the Spring Festival Gala project
It is reported that more than 3,000 technical personnel in the Jingdong technical system have participated in the technical research and support of the project, and as many as 2,000 technical support personnel participated in the front-line duty on the day of Chinese New Year's Eve. Behind the more than 10,000 Jingdong people working together is the wonderful moments such as the New Year Goods Festival and the Spring Festival Gala Red Envelope Interaction spanning up to a month, and more importantly, it has created the ultimate experience for consumers across the country and led hundreds of thousands of merchants, customers and partners to share the festive and peaceful feast.
Use the "Transformers" thinking to deal with this "more difficult"
Fingers crossed, the Spring Festival Gala red envelope interaction has entered the seventh year, this year Jingdong Cloud in the first escort tiger Year Spring Festival Gala completely bid farewell to the traditional operation of simply increasing server resources to overcome high concurrency, with more efficient and agile resource transfer scheduling and cloud native infrastructure, hybrid multi-cloud operating system cloud ship and other technical killers, according to the "battlefield" changes and rapid "transformation" to meet the needs, just like Transformers.
In this regard, Chang Liang, head of the basic guarantee of the 2022 Spring Festival Gala Project IDC, senior director of the Jingdong Cloud Infrastructure R&D Department, and senior director of the JD Cloud Infrastructure R&D Department, said: "In order to cope with the interaction of the Spring Festival Gala, we did not prepare additional resources separately, on the one hand, because the short-term temporary investment was too large, contrary to the refined R&D resource management concept pursued by JD Cloud for a long time; on the other hand, due to the global supply chain tension caused by the epidemic, the objective path of increasing resources became unworkable. So just through the previous '618' and '11.11' resources, through internal rapid movement and expansion, to achieve second-level scheduling of nearly 3 million containers, more than 10 million accounting resources in the Spring Festival Gala interaction and the Spring Festival of the Spring Festival in the two modes of rapid switching, although the difficulty is great, but we succeeded. ”
Chinese New Year's Eve the Jingdong cloud technicians on duty to deploy the work before the Spring Festival Gala
From borrowing resources to meet complex needs, transitioning to improving the efficient and agile ability of the system architecture to quickly change the array to cope with various challenges, especially forging the ability to enhance the extreme concurrency of large-scale scenes, to some extent, represents the continuous innovation of cloud manufacturers' technical capabilities.
Digging deeper from the perspective of resource optimization, we know that the difficulty of the spring festival red packet interaction lies in the full-link complexity brought about by the superposition of "red packet + consumption", "under such a high concurrent traffic, the double-active architecture used in the past in the promotion link is obviously insufficient to cope with the challenge, so it is very important to complete the system classification through business evaluation to achieve the optimal allocation of resources." Based on this, the project preparation team anticipated and developed a Grading Standard (SABC) to complete the dynamic adjustment and renewal of resources. For example, in the interactive session of the Spring Festival Gala, the red envelope interactive link system is S-level, and others will be downgraded as appropriate, so as to ensure that the high-priority application system in the interactive process gets as many resource use opportunities as possible, so that "less resources to do big things".
Of course, in terms of agile and flexible resource response and improvement, the Spring Festival Gala red envelope interactive project routinely showed the "ace of the house" of Jingdong Cloud, that is, a powerful hybrid multi-cloud operating system cloud ship. Thanks to the stable support of cloud native technology from within, JD Cloud ensures the optimal global resource orchestration and cost and the most stable system operation by exerting its agile scheduling for large-scale heterogeneous infrastructure and flexible and smooth expansion of resources. In particular, the embedded intelligent scheduling system uses machine learning and deep learning intelligent algorithms to predict the use of application resources to complete elastic optimization; at the same time, the ultra-large-scale off-line mixing technology is also in the preparation of this technology to resolve the long-term pulse traffic peak to achieve the full use of limited resources, the original three machines can solve the problem, this time as long as one is done, so that the computing power can play the maximum value.
When it comes to making "human calculation" to the extreme to meet the test of "heavenly calculation", the project team also has a lot of experience in this preparation. To achieve flexible and agile response, the team predicted the possible traffic distribution before the event. For example, based on the multi-dimensional data of watching the Spring Festival Gala and participating in the activities in the past, the "traffic map" was first drawn to predict the difference in regional traffic in advance and carry out targeted deployment of resources.
Regarding the prediction, Zhang Jinzhu, head of the Spring Festival Gala project T-PaaS and middleware, made a vivid analogy: "If the red envelope interactive project is seen as allowing hundreds of millions of spectators to quickly enter a certain venue to watch the game, the middleware is actually equivalent to the various channels of the venue." We need to cooperate with the resource scheduling system to quickly and reasonably open these channels, undertake the influx of people, and ensure that everyone enters their seats in an orderly and controllable manner to watch the game, in fact, to ensure that everyone can smoothly participate in the red envelope interaction. In this regard, we will prejudge the possible links of large traffic according to the data analysis we currently have, and do a good job in the deployment and adjustment of corresponding resources, prejudge the path and direction of traffic flow, do a good job of controlling and guiding timely response to resource needs to expand and shrink, and ensure that 'limited resources are used on the blade' and quickly a word. ”
The interaction of the Spring Festival Gala and the rapid landing of the guarantee plan, in addition to reflecting jd.com's cloud building block IT thinking to a large extent, are also the normalized preparation experience of coping with their own 6.18 and 11.11 for many years. This time, in response to possible unexpected situations, JD Cloud pioneered the "emergency script" as an abnormal drill plan. "In the process of preparing for the battle, through up to 7 rounds of stress tests, including public network stress tests and disconnection drills, etc., repeatedly practice the operation steps of the plan and observe the effect, and at the same time examine the health of the application running on the system, and then verify whether it meets the expected effect and continuously adjust, better respond to sudden business abnormalities and module abnormalities, etc., and escort the interaction." The re-insurer concluded.
Today's JD Cloud can help R&D to carry out full-link, all-round architecture upgrades and refined resource management through a one-stop safe and efficient production system, so that more and more systems are relied on to ensure the stability of major nodes, rather than relying solely on technical manpower, which is the valuable experience of the Taishan project polished for two years, and it is also the ability to quickly and smoothly cope with special business scenarios.
Peak flood superimposed fulfillment from simply going to the cloud to better use the cloud
Behind the flood of traffic at the Spring Festival Gala, which is still fresh in memory, is a huge and complex world-class supply chain application scenario. As we all know, this time Jingdong in support of the Spring Festival Gala at the same time shouldered to support the national "Spring Festival" in the overall supply chain performance of retail and logistics, involving front-end App platform, orders, settlement, payment, search, recommendation, to the back-end warehousing, distribution, customer service, after-sales and other business systems, it can be said that relying on years in the business scene to refine the "cloud chain integration" high response, high agility ability, successfully sang a song: from simple "cloud" to meet business needs to "use the cloud", The song of improving innovation efficiency not only shows its own hard-core technical strength, but also can be seen as a silhouette of the innovative development of China's cloud computing industry.
For a long time, Jingdong has carried out application innovation on the whole link of the supply chain, realizing the world-class inventory turnover of nearly 10 million self-operated goods, and the minute-level delivery of more than 300 cities across the country; using the intelligent supply chain super automation to complete the intelligent decision-making + automatic purchase of the whole process of goods, and adopting the "Jinghui" digital supply chain service solution to provide integrated supply chain optimization decision-making services for massive merchants, and this time it is to join hands with more entity enterprises to continue to achieve high-quality growth during the Spring Festival.
In addition, in 2022, as the 10th consecutive year of Jingdong's "Spring Festival Delivery": let consumers in 30 provinces, more than 300 cities, and nearly 1,500 districts and counties across the country, even in Chinese New Year's Eve and the first day of the Chinese New Year, can place orders and receive goods normally, enjoy the Spring Festival logistics services delivered by more than 200,000 Jingdong logistics brothers, and fully meet the service needs of merchants and consumers during the Spring Festival with the "Spring Festival of New Year Goods".
Of course, in order to achieve the "faster and better" logistics distribution goal and ensure the short-term efficient and accurate delivery of orders, the platform's intelligent order distribution system, capacity control system, and real-time synchronization of two-way data between customers and merchants are facing great challenges, and the safe and stable cloud service base is particularly important.
At present, Jingdong Logistics has achieved full cloud, "running" on the cloud database according to the pre-estimated amount of data to do a good job of resource planning and allocation, the real peak of the cloud database will be through a series of database-level technical means such as high availability architecture, automatic failover, elastic expansion mechanism, etc., to ensure that the data can be backed up, fault can be switched, incremental can be expanded, calmly cope with the data pressure during the peak traffic period, and fully realize the full link guarantee for Jingdong's "Spring Festival" Spring Festival for 24 hours.
JD Cloud supports JD.com to deliver goods for the 10th consecutive Spring Festival
"This year's Spring Festival Gala is also a very big challenge for the payment side", the person in charge of the payment end of the Spring Festival Gala project said: "In addition to the traffic on the original order transaction link this year, multiple links such as prize issuance, red envelope viewing, and user login registration will be tested by the traffic peak." In this regard, the payment and settlement service platform of JD Cloud's independent property rights makes full use of the dynamic rule splitting algorithm patent to solve the problem of massive data reconciliation, and achieves a continuous uninterrupted capacity of the core trading system of more than 99.99%, with strong disaster tolerance capacity and considerable throughput, successfully coping with the peak traffic of the Spring Festival Gala while providing users with a safe, stable and convenient red envelope interactive experience.
In order to ensure the user experience of consumers during the Spring Festival, Jingdong's first intelligent digital customer service Qianyan is also officially on the job, committed to providing a technical and temperature customer service experience; behind the natural communication and smooth interaction of Qianyan is the Jingdong Intelligent Customer Service Yanxi independently developed by JD Cloud to provide the world's leading multimodal interaction technology. During the Spring Festival, Qianyan and other "customer service partners" will carry the highest SaaS service traffic peak in history, providing users with 7*24 hours of full-link scenario intelligent services. According to the latest data, since the start of the New Year Goods Festival to the end of the Spring Festival Gala, Jingdong intelligent customer service has accumulated 550 million consultation services, providing more than 100 million services for 165,000 merchants, accompanying the people of the whole country to comfortably handle New Year goods and spend the New Year with peace of mind.
After the 360-degree test of the shortest preparation time in history, the longest interactive activity in the history of interactive activities, the world's largest network interactive activity, and the world's most complex scene, for JD Cloud, in addition to the comprehensive technical ability improvement of the Spring Festival Gala Red Packet Interaction to cope with extreme and complex scenarios, it is also a deepening from traffic competition to industrial digitalization, from business "going to the cloud" to "using the cloud better".