(Yicai Global) Oct. 30 -- Singles' Day, the online Chinese shopping festival celebrated annually on Nov. 11, has led to a technology revolution. Traditional financial transaction systems aren't efficient enough to handle the sales boom, forcing engineers to push the envelope with mobile payment solutions, Renwu said in an article published on social media today.
During the event, e-commerce behemoth Alibaba Group Holding Ltd. [NYSE:BABA] organizes for sellers on its online retail platforms to offer up discounts.
Both visitor traffic and trading volume on the company's Taobao and Tmall platforms treble at midnight on Nov. 10, with both figures skyrocketing every year. More than 120,000 people placed orders on the sites at 0.00 a.m. on Nov. 11 last year, paying for more than CNY1 billion (USD150 million) worth of goods just 52 seconds later. As the day closed, people from over 200 countries and regions across the world had taken part, spending CNY120.7 billion (USD18.29 billion), said Hu Xi, chief technology architect at Ant Financial Services Group, Alibaba's financial arm.
Proficient buyers on the two platforms can complete their transactions in as little as three seconds. The technological developments that enable that have brought some major changes to the world.
Hu says he hasn't partaken in Singles' Day as a buyer for the past nine years.
For years, the festival would send him into an uncontrollable state of anxiety, he said. That feeling would spread across the entire office until the night of Nov. 10, when programmers at the firm worship Guan Yu, an ancient Chinese general seen as the epitome of loyalty and righteousness, with cigarettes, milk, fruit and red underwear, which is good luck in traditional Chinese culture. Staff would print out his portrait and place it on a table at the center of the office, overwhelmed with anxiety and fear. Nobody remembers why they chose to worship Guan Yu, "it's just a sign of our reverence for technology," Hu said.
The technology behind the event, or rather, a potential collapse of the technology, is awe-inspiring. Singles' Day may be a day that netizens flock to purchase discount goods, but it's also a major source of problems for Alipay, Alibaba's mobile payment platform, which struggles to contend with the vast quantity of transactions.
This technological 'black hole' began in 2009, when the first Singles' Day festival was held. Back then, people had no idea of what was to come. It was planned as an online marketing campaign to boost sales for lesser-known Taobao stores. Only a dozen or so employees worked overtime that first night, with technical staff leaving the office as usual. There were no press conferences, no warm-up sessions, and no countdown or celebrity endorsement -- just a few employees making preparations for the midnight promotion.
Shortly after midnight, sellers' printers were catching fire due to the sheer volume of orders and Alibaba headquarters was pulling in security guards to help keep up with the unprecedented volume of inbound phone calls.
After massively underestimating the initial campaign, the firm doubled down on its backup capacity the next year, assuming that would be enough. Even if midnight sales did shoot up, buyers would have calmed down by later in the morning.
However, the e-commerce giant once again underestimated Chinese shoppers. Alipay transactions tripled right after midnight and volumes were at peak levels throughout the day, forcing Cheng Li, chief technology officer at Ant Financial, to order his team to divert resources away from other business to the core transaction system. Programmers were shutting down other functions until the last 30 seconds of Nov. 11, at which point the system was seconds from collapse.
Nobody had forecast the incredible growth the event experienced. After Cheng joined Alibaba, he devised an eight-digit transaction code system designed to accommodate up to 99,999,999 transactions. He was confident in the system, boasting that the numbers wouldn't be used up in his lifetime.
On the eve of Singles' Day 2012, the system was just about done. The firm has since upgraded to an 11-digit system, but nobody knows for sure how long this will hold out.
Midnight bargain hunters are difficult to keep happy. Hu's calculations show that even patient users wait only 30 seconds after a system failure, any longer than that and they'll abandon the order. On Singles' Day, some shoppers can be agitated by as little as a 0.1-second interruption, he said, which can prove fatal for apps relying on user retention.
All shoppers may place orders or initiate payment transactions at midnight of Nov. 11 -- just imagine the large crowds of customers rushing to the counters at a shopping mall when it opens for a midnight sale.
Alibaba's transaction statistics show that the peak value of simultaneous trading volume at Singles' Day midnight has continued to increase by the year, from 500 in 2010 to about 1,500 in 2011 and to a staggering 120,000 last year. A peak value of 500 means that up to 500 payment transactions can be successfully processed and settled per second, and the 501st and later transactions will go to the waiting queue.
Consumers' ever-growing spending desires mean that the e-commerce platforms must constantly upgrade their transaction systems to meet users' insatiable desires.
The most obvious solution is to create more 'counters.' The 'IOE' (IBM minicomputers, Oracle databases and EMC storage devices) architecture is a powerful centralized point-to-point framework that is mostly commonly used in traditional financial services. It works like an expensive, reliable and robust centralized 'counter' but has limited processing capacity. To cater for more customers, the company must deploy more 'centralized counters' that cost more than USD5 million each. This would be an astronomical bill for massive e-commerce platforms like Alibaba.
In addition to the prohibitive cost, the IOE solution entails a colossal waste of resources, because the 'centralized counters' are completely useless after the Nov. 11 shopping season. "We can't manage the spike in transaction volume on the Singles' Day, but they will be lying idle after Nov. 11."
Even worse, the high degree of centralization of the IOE architecture means that all important accounts are stored in one place, and a single system breakdown will result in all customers having to abandon their orders.
The phenomenal success of Singles' Day sales prompted Alibaba to phase out the IOE model, and pursue a different approach. The firm has migrated all transaction processing operations on to a cloud computing platform, and continuously optimized the performance of the new system. They have developed a new solution, dubbed distributed 'parallel redundancy' flexible architecture.
The new backup solution is developed based on the following premises: deploying the most common personal computer servers to satisfy the minimum data storage hardware requirements, and applying the most robust logic to meet the maximum reliability requirements in financial transactions. By analogy, the 'parallel redundancy' architecture replaces the centralized counters with innumerous small computers installed throughout the 'shopping mall.' Their computing capacities are much smaller than the central counters, but every computer can process a certain amount of transaction. More importantly, they are coordinated by a cohesive force -- cloud computing, which is placed at the core of the hardware and software architecture, that allows the computers to serve as backups for each other, thereby ensuring that all the computers function effectively as a whole, and that a system failure at an individual computer does not affect the overall performance.
It is Alibaba's solution to the Singles' Day challenge, which is probably the most advanced solution in the world today, Hu opined. "Developing the distributed (architecture) poses a serious challenge, because it must be a distributed system and compatible at the same time. Furthermore, the new system must be able to deal with very challenging tasks, and we needed to use offline resources while ensuring compatibility, which involves combining totally incommensurable resources and creating the highest level of online capacity," he explained, "I feel that our job is to continuously set new records in terms of improving technology's capability in responding to new challenges."
In the ever-evolving technology industry, people's growing demand and curiosity have stimulated technology development beyond its previously perceived limits. Alipay must have acquired the capability to process 160,000 online transactions per second last year, Hu estimated. His team has managed to "increase the system's capacity by twice the size that they did last year" with no extra cost.
The selection of such a unique architecture was a bold decision at that time. We are probably the first financial service firm in the world to deploy core banking systems on a distributed cloud computing platform, he noted. They are a pioneer of an emerging market trend. At present, most internet companies have abandoned the centralized IOE architecture, and China's banking regulator have mandated major banks to shift away from the IOE model toward cloud computing platforms.
Hu's efforts have paid off. When operating in the same transaction environment, traditional banks that use the IOE framework pay approximately CNY1 (USD0.15) per transaction in processing costs, but Alipay has trimmed the cost to CNY0.02 to CNY0.03.
Technology teams at Alibaba are taking even bolder actions to speed up technical innovations in line with the growth in consumer spending. Hu is more interested in the 'next-generation fintech diagram,' he says. The first-generation fintech generally focuses on channel related innovations -- "how to move counter-based operations to the internet." Later technologies seek to "put everything on the internet." Hu's work now goes well beyond 'technology infrastructure development' and also covers risk control, credit and user connection.
Having witnessed the technical revolution driven by the consumer boom first hand, Hu has changed. He will have a special Singles' Day this year, the chief technology architect said. Some of the biggest challenges that his team tackled over the years have now become an everyday occurrence. In the past, he had to assemble 300 to 400 people on standby for Singles' Day operations, but there will only be around 20 team members working on the night of Nov. 11 this year.
At last, the technology chief at China's largest e-commerce company can have time to enjoy the largest online shopping festival in the country. For the first time, he loaded his Taobao shopping cart with various products that he plans to buy at discounted prices on Singles' Day. He will buy decorations for his newly furbished apartment and Lego toys for his daughter. He may even try his luck with midnight sales of premium Chinese liquor products.