Shulou(Shulou.com)06/03 Report--

Zhang Wensong: Challenge difficult problems and do more valuable things with technology

Why did you choose IT? Most people would probably say, because of the money. Yes, it has become an indisputable fact that this industry has good treatment and high salary.

But for Zhang Wensong, whether it was creating an open source project LVS during college, working for Alibaba for seven years, or rejecting Google and Facebook, last year ×× Didi Chuxing. All adhere to an initial heart: challenge difficult problems, use technology to do more valuable things

Zhang wenzhao

senior vice president of Didi Chuxing

Zhang Wensong, founder of Linux Cluster Project-LVS. He is currently the senior vice president of Didi Chuxing, responsible for the basic platform work and the overall person in charge of the intelligent transportation project. He has worked in Alibaba for nearly seven years, successively serving as Senior Technical Director of Taobao, Chairman of Taobao Technical Committee, Vice President of Alibaba, Chairman of Alibaba Open Source Committee, Alibaba Cloud CTO, etc.

Accidentally, the "seed of open source" sprouted

In 1995, Zhang Wensong also studied for a doctor at the University of National Defense Technology. He found that the Load Balancer equipment used in Changsha's construction of an information port was extremely expensive. At that time, I had a bold idea-could I do Load Balancer on Linux?

Just do it! It took only two weeks to complete the code of the open source software LVS. At that time, there was no content related to Load Balancer in Linux field, and this project was very practical, so after putting it on the Internet, someone responded quickly. Not only are there requests, but there are also feature patches from contributors. LVS projects iterate rapidly and become more and more functional.

This experience enabled Zhang Wensong to communicate with the world's top technical talents and experience the charm of open source. Therefore, when he joined Taobao later, he always advocated open source and created a low-carbon network through technological innovation.

If the LVS project is a point

So Taobao is a face

Join the more challenging Taobao, Zhang Wensong single-handedly set up the core system team, corresponding to Taobao users of many basic services, support the whole business.

The brand-new system planning should start from hardware customization, to operating system, and then to stand-alone software optimization. Such as Tengine, JVM and database optimization.

Among them, the database is the biggest bottleneck of the whole trading platform. How to deal with 100,000 transactions per second from Taobao business, CDN system, distributed storage, distributed cache system, big data platform, image search, etc., but these are only a small part of Zhang Wensong's experience in Ali.

Taobao as the largest online shopping platform in China, the whole process of online shopping is selected, make the final decision, at most in the selection process to make recommendations or advertisements. How do you use technology to do something more valuable?

Artificial intelligence has come, and new challenges have come.

This time last year, Zhang Wensong ××× Didi. Didi's transformation in the field of transportation has begun.

When we use Didi, we just need to enter a destination and we can rest easy. But behind this seemingly simple UI, all rely on calculation to complete, and how to achieve perfect and effective matching is a very complex calculation problem, but also requires a very high real-time performance. All of this uses artificial intelligence algorithms.

"Didi Traffic Information Platform" Beijing Road Traffic Thermal Map

road condition prediction

Based on a large number of real-time reported trajectory data, real-time road conditions can be calculated, and future road conditions can be predicted according to historical road condition data and real-time road conditions.

path planning

Route planning is done for the starting point and destination on the whole road network, including the receiving end route and the sending end route.

ETA calculation

ETA technology incorporating a lot of machine learning is used to predict the time of receiving and sending off both ends.

smart distribution

The ideal dispatch match is not to send the nearest driver to pick up the order when the passenger issues the order, which may happen a millisecond later when another passenger's demand is more consistent with the driver's demand. In view of this situation, first put the passenger billing demand into the demand pool, Didi in order to take into account the user experience, set the matchmaking time to 2 seconds/time.

Intelligent sorting constructs a huge supply and demand matrix, with a list of passenger order requirements vertically and a large number of available drivers horizontally. The platform needs to calculate the correlation between users and drivers.

There are many dimensions that can be referenced by correlation, such as physical location, mapping of location to road network, driving time, driver service score, passenger preference, etc. These multidimensional terms are weighted and mapped to the space between 0 and 1. The larger the value, the higher the correlation between passengers and drivers. This process should be controlled within 100 milliseconds. Because no-load drivers in non-stop driving, position changes at any time, real-time is a factor that must be improved.

In addition to this, there is the imbalance between supply and demand in different areas of the city. For example, if there is a demand from passengers A and B, the driver is closer to A and slightly farther away from B. It is very likely that the driver will be assigned to A by 2 seconds. However, if destination A is oversupply and destination B is in short supply, it may be more reasonable to send drivers to B first from the perspective of overall supply and demand balance.

Do something more valuable with technology_intelligent transportation

As the whole online car-hailing platform, due to the business characteristics of route settlement price, it is necessary to report GPS positioning point data to the platform in real time, and these positioning points are connected to the driving route. Millions of cars a day, driving N routes, can well reflect China's urban traffic conditions, this is a new direction that can bring social value.

In the past, most of the traffic systems used detectors, coils, bayonet, cameras and other ways to collect data at each intersection, with long cycles and poor timeliness.

Now, based on Didi's massive data, it is used for traffic management, operation, supervision and decision-making, recommending the best route to consumers and bringing more convenience to users 'travel.

Didi will contribute to intelligent traffic governance, intelligent operation, overall traffic measurement system construction, intelligent travel of all travelers, etc. Relying on its advantages in several aspects of "blocking" in the big intelligent cloud (Internet of Vehicles platform-mobile Internet-big data platform-artificial intelligence algorithm-cloud computing platform), it brings changes to the whole transportation field.

For example, the Didi Smart Signal Project is being implemented, and the signal lights are adjusted and optimized according to the changes in traffic flow.

Jinan Jingshi Road 6 intersection signal lights optimized, delay time reduced by 10%

Signal lights at 6 intersections of Jinan Weishier Road are optimized, and the delay time is reduced by more than 15%.

This is just the beginning. Didi currently has more than 20 million daily orders, but in the entire travel sector, this number is not large, accounting for only about 1.5% of all trips.

From the perspective of inclusive travel, how to do valuable things with cutting-edge technical means, how to make public transportation more stable and predictable, how to accurately recommend the optimal transfer method of various transportation vehicles in real time, etc. are all the future directions of Didi.

When it comes to dreams and feelings, Zhang Wensong said that from teaching in university, to leaving to start a business, to Ali and to Didi now, in terms of personal personality, it feels better to do truly useful and valuable products. What to do will start from the perspective of solving problems for people and bringing value to society. In this process, we constantly make choices about key problems and solve problems continuously, which is also a process of continuous learning and continuous improvement for ourselves.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.