Tech creator·StreamNative Zhai Jia: The value of the open source model is king, and the future of basic software lies in the domestic community

Zhai Jia, co-founder of

Member of Apache Pulsar and Apache BookKeeper PMC, former technical leader of EMC unified storage department, former founding engineer of Streamlio. He was selected as the "Top 33 Open Source Pioneers in China" in 2020, and was awarded the title of "OSCAR Pinnacle Open Source Figure" in 2021, an open source technology evangelist.

—
Text | babayage
Edit | smile

Innate preference for bottom principles and long-termism
First acquainted with open source, believe in value as king

With a little contact with Zhai Jia, you can feel a sense of determination and solidity from the heart. Friends commented on him as "the unity of knowledge and action, believe in the underlying principles and logic, and do it if you believe it."

Speaking of why he embarked on the technical route of storage, Zhai Jia frankly said that “there is no dream, mission is so lofty.” He insisted on finding a reason, probably because of his willingness to delve into the bottom layer and study basic theories. : Undergraduate majoring in computer system architecture, there are not many options before the postgraduate period: choose one of communication or storage. After graduation, it is logical to add EMC (Easy Security), a synonym for information storage. At that time, the mobile Internet wave hadn’t rushed forward, and there were not many forward-looking people who could predict that storage would become an important infrastructure in the digital age. A thing worthy of long-term research and learning."

During EMC, Zhai Jia came into contact with BookKeeper, and thus ushered in the second key word of her career: open source. Speaking of the initial feelings about open source, "it must be very different from now. The initial understanding of open source is public code, which everyone can obtain and use, and it has been used in some more advanced company scenarios and has been tested in practice. They are all excellent and reliable choices."

Every hardcore player has a little white history, and Zhai Jia's deep understanding of open source also comes from years of accumulation and iteration. But what is different is that since the day when he came into contact with open source, he has always adhered to the concept of "value is king". He is not too cold with all the concepts, cultures and declarations that put the value of the scene behind. "The value of software lies in solving the world. There is a certain problem in the above, open source is just an effective means to create value."

One of those pitfalls of entrepreneurship:
Choose the wrong model, choose the wrong community

Looking back on Zhai Jia’s entrepreneurial career, it’s difficult to bypass his friend: StreamNative CEO Guo Sijie, both of whom are graduate students at the Institute of Computer Science of the Chinese Academy of Sciences, are also deeply involved in the field of storage technology and participate in the BookKeeper open source community... a proper official CP is both Sense of sight.

In 2016, Zhai Jia resigned from EMC and started a small house model for half a year. At this time, Guo Sijie has embarked on the road of entrepreneurship and founded Streamlio around the end-to-end solution of "Heron+Pulsar+BookKeeper". He heard that his friends were relaxing at home and invited him to join. Zhai Jia felt that "the underlying logic is very reliable. The value is obvious, so try it out."

However, in the more than a year after Streamlio was founded, they have tried various commercialization methods with little success. The founder team gradually formed a consensus: the commercial promotion of basic software requires a lot of resources, and it is a giant game. ; Moreover, lack of basic software that demonstrates the value of the scene, no matter how advanced the technology is, it is difficult to be recognized by the market.

As a result, the open source community became the Huashan Road for Streamlio PlayerUnknown's Battlegrounds, but they made another minor mistake in the selection of the open source community: first they chose the American community with the most developed open source community. Zhai Jia reflected on this. : "Enterprises that do basic software entrepreneurship are actually more suitable for an environment with fast development, big changes, and rich data soil-that is, domestic communities."

After a pause, Zhai Jia said one word at a time: "For entrepreneurial projects that follow the open source community route, we must fully consider all factors and choose the community carefully, because except for a few pioneering technology projects For most projects, the community plays the role of productivity, demand pool, promotion channel and moat at the same time-without a community, there is no life."

In October 2018, Zhai Jia and Guo Sijie held two Meetups with the theme of Pulsar in Beijing and Shanghai to test the operation, construction and promotion of the community, and the response was far more enthusiastic than expected.

In the same year, the two left Streamlio to create StreamNative to create a basic software entrepreneurial model based on the domestic open source community.
Technologist: More and more entrepreneurial projects choose the open source model. Can you share the criteria for companies to choose a community?
Zhai Jia: This topic is a bit bigger. The only thing I can clearly suggest is that companies doing basic software projects may wish to try the domestic community, because the business model of domestic companies is still driven by business, with rapid growth, large changes, and scenarios. Rich, as long as your product is good enough and efficient enough to solve business problems, they will be willing to use it. In addition, in fact, there is no essential difference between community marketing and To C marketing. WeChat's marketing target is social demanders, and the marketing target of open source software is programmers. The essence is the connection between value and people.

Two signs of "seeing light":
Landing in the big factory scene, Meetup is full

I have witnessed that BookKeeper, which has excellent technical principles, is difficult to commercialize because it is too abstract. I have experienced the Streamlio period holding its breath under the water, and I don’t know when to breathe. I have experienced what I think is a subtle end-to-end solution in the American open source community. After a cold encounter... After a series of difficulties and setbacks, Zhai Jia still firmly believes in the technical direction of cloud native storage. He believes: "The environment, timing, decision-making, and model may all be wrong, but this direction must be correct."
And his persistence finally saw the rewards. With full trial and error and iterative experience as fuel, StreamNative finally set sail smoothly. Eight months after its creation, StreamNative held a Meetup. Zhai Jia still remembers the scene of that day: from 9:30 in the morning to 6:00 in the afternoon, the participants included domestic partners from major manufacturers such as Tencent, and others from Yahoo! Japan’s overseas partners specially flown in, “I found out when it was about to end that we had been able to hold up a whole day of Meetup. That day, I felt from the bottom of my heart that this career has come to light.”

More direct confidence comes from the affirmation of Beacon customers. Tencent's billing platform and short video application BIGO have respectively become beacon cases for StreamNative's online business scenarios and offline data analysis business scenarios:

Tencent’s billing platform not only has requirements for system expansion, but also has strict requirements on data service quality. Tencent’s billing platform uses Apache Pulsar to process an average of 10 billion+ transaction requests per day, and consumes 10T+ data per day, which carries Tencent Group’s daily data. 100 million in revenue, and the total number of custody accounts reached more than 30 billion. The BIGO case is based on the pain point of cluster operation and maintenance that everyone often encounters in streaming scenarios. BIGO builds real-time recommendations with the help of Apache Pulsar and the good integration of the big data ecosystem And the analysis system helps the rapid development of the business and reduces the cost and difficulty of the original Kafka cluster operation and maintenance, especially the labor cost of expansion and contraction.

Zhai Jia expressed his sincere gratitude to the early customers. In his view, "The essence of cooperation is to reach a consensus. For the basic software to be truly successful, a large-scale and large-scale consensus must be reached, otherwise, what is the "foundation"? Because of this, the cost and difficulty of commercial promotion of basic software is extremely high. Through the open source community, it is the best credibility and the advantage of the open source model to quickly realize the value of high trust and difficult scenarios."

Three or four of those pitfalls of entrepreneurship;
Push people, misjudge the priority of requirements

Technologist: After starting a business, what experiences have you had to subvert your past cognition?
Zhai Jia: A lot, I can say a few casually (laughs).

The first is being pushed away by things. Many start-ups have gone through the stage with fewer people and more tasks. When they get busy, they just focus on "work", often forgetting what they are doing, and they don't care about talent training.

But if there are too many pitfalls, you will know that this is not feasible. There are two solutions: one is to establish a plan and clarify the action goal. Talent, what kind of people do we need to do this? If you think about it, you can find someone; the second is to be fully authorized, if you don’t have enough people, you must trust your comrade-in-arms.

The plan has been made, and the new big hole has also come. The most common problem in the entrepreneurial stage: resources are limited, which is the higher priority of customer needs and strategic planning? At first we thought that it was correct to meet the needs of paying customers. The customer is God and people should work for people with money, but in the end we found that this was wrong. There is a fundamental problem with the model of "customers raise demand-we meet". : The needs of customers may not be the best solution to the problem. The best solution is more likely to be in our strategic planning, because we are the technical experts in this field and understand its existing problems and future potential better. Therefore, we have increased the process of communicating priorities with customers to help customers understand that "problems can be resolved more effectively in accordance with our plan."

Facts have proved that customers choose you because they trust you, and those who trust you are willing to trust your judgment, as long as the communication is clear.

Technologists: "Technologists" has shared a lot of the authorization experience of predecessors in science and technology. Would you please share the practical experience of your authorization?
Zhai Jia: My style is firm or even radical authorization. I fully trust my colleagues. Before implementation, I am only responsible for determining plans and aligning goals with colleagues. After that, I don’t care about specific things. I encourage colleagues to do boldly and go with confidence. Problems arise. I'll help you review the situation again, think about how to solve it, and keep doing it if there is no problem.

The future of basic software in China

Technologist: In the course of StreamNative's entrepreneurship, has there been any strategic swings?
Zhai Jia: Yes, but not at the technical level, but at the market strategy level. StreamNative has a group of overseas users whose willingness to pay and usage scenarios are different from those in China. They are more accepting of the SaaS charging model, while domestic users need to solve specific problems and tend to privatize deployment. At this time, there will be many different voices. For example, someone will suggest that you can already acquire customers through the SaaS model overseas. Is there any need for these domestic customers?

StreamNative has carefully considered the path of "overseas market + commercialized SaaS", but finally decided to stick to its value proposition: The most important thing is not which model makes more money and makes money fast, but which model and product's future is truly true Combine it together-with rich domestic customer scenarios, large amounts of data, and rapid development, StreamNative will polish its products with the scenarios of domestic communities and users, and feed back SaaS products.

Of course, it cannot be denied that the willingness of foreign users to pay is indeed much better than that of domestic users. It has also given us a lot of encouragement, but StreamNative hopes to see the domestic environment continue to improve. Now, the national level is paying more and more attention to open source, and StreamNative will also Participated in the overall digitalization process in China.

Future planning: unified product line

Technologist: Tell me about the planning and goals of the next stage?
Zhai Jia: The core work is to achieve the unification of the entire product line. We hope that all products use the same framework and code.

In the overall direction, it is still necessary to control the advantages in the two directions of message (MQ) and big data flow processing. One direction we are doing recently is to get closer to the big data pipeline scene. We are discussing with the Flink community how to apply Pulsar in batch-stream integration scenarios, and give full play to the advantages of Pulsar's cloud-native architecture that separates storage and computing.

In the message (MQ) direction, everyone pays more attention to the analysis of various message protocols. We are also working hard with other friends in the community to enrich this part of the ecology.