From these cloud-native companies, I see what a digital innovator should look like

01

In 2020,

The Chinese market has opened the first year of the large-scale implementation of cloud native.

Since then, cloud native has widely entered the vision of non-Internet enterprises.

Cloud native architecture has gradually become the mainstream way of application deployment in the past two years.

To further develop the digital transformation of enterprises and achieve scale digitization,

Modern application forms with elasticity, distribution and continuous evolution,

It is an important way and means for enterprises to realize digital innovation.

What will the digital world of the future look like?

How will enterprises with cloud native maps subvert digitalization?

With these questions, we interviewed

Many big names in education, finance, new retail, mutual entertainment, media and other fields,

They are leaders in digital innovation,

He is also a practitioner of creating business value with cloud native technology.

Next, let's listen to

How have they changed by being "cloud native"?

homework help

Cost reduction and efficiency increase is a major requirement of the company's infrastructure. On the application side, it is necessary to increase the unit computing power carrying capacity, which is generally referred to as QPS. But one of the challenges we face is that the homework help technology stack is too diverse. How to improve QPS as a whole? Looking at the resource side, resources such as storage and network are either just needed, or it is difficult to control costs. The focus of cost reduction on the resource side is computing resources, and for computing resources, we need to increase the computing power per unit cost.

"In my opinion, in the process of reducing costs and increasing efficiency, we should pay special attention to reducing costs and not reducing quality . When reducing costs, stability, efficiency, and security cannot be compromised. We choose to work with Alibaba Cloud, and choose the power of open source. Combined with certain self-research to solve related problems. At the application level, we have improved the operating performance of the mainstream technology stack, and reconstructed the architecture of the most used retrieval services to improve performance and operation and maintenance efficiency.

In the separation of computing and storage, we introduce Fluid as a key link. Fluid is a K8s-based data orchestration system, which is used to solve a series of problems encountered in the cloud native process, such as complex data access process and slow data access. JindoRuntime is used to achieve cache acceleration. When we use Fliud and JindoRuntime to complete the entire process After the reconstruction of the retrieval system, the benefits obtained are also more obvious. The data update cycle of Jobbang has been shortened from the previous hour level to less than three minutes, and the delivery of the entire machine for operation and maintenance has been shortened from the previous day level to the hour level, and the program performance has been improved by 30%, resulting in a reduction of ten thousand core level resources.

In the future, Homework Gang will migrate more online services to ECI to achieve real peak shaving and more cost-effective IaaS resources, which is also the direction we have been trying and exploring. "

NetEase Cloud Music

Most of NetEase Cloud Music's original audio and video technologies are applied to the data processing of the music library. The cloud music music library team and the audio and video algorithm team jointly built the NetEase Cloud Music audio and video algorithm processing platform, and implemented more than 60 kinds of audio and video algorithms. , especially in the past year, service-based algorithms have accounted for half, and these algorithms have provided service capabilities to cloud music 100+ business scenarios. However, more complex algorithms and more business scenarios have put forward higher requirements for NetEase Cloud Music's service efficiency, operation and maintenance deployment, and elastic capabilities. Before we went to the cloud, we had used more than 1,000 different devices internally Specifications of cloud hosts and physical machines.

"We quickly tried the function computing FC within a week, but a complete and highly reliable architecture needs to consider more factors. Therefore, our transformation focus is to pop out the computing power task through the function computing FC, and the system is in the overall External input and output remain unchanged.

In the first stage of introducing function computing, the algorithm of feature extraction has been improved by 10 times ; the algorithm of sparse class has been greatly saved in cost. In addition, through the image cache acceleration capability of function computing, the startup speed of the node is optimized, so that all services can be pulled up in seconds. These works have reduced the operation and maintenance cost of algorithm operation and maintenance processing, allowing us to focus more on the algorithm and the business itself.

In the future, we hope to further liberate our human investment in operation and maintenance through serverless technology, and will try to store audio and video algorithms in more scenarios. "

Amway

Under the impact of the Internet, Amway not only needs to adapt to the core requirements of agile, fast, and low-cost trial and error in the Internet scenario, but also needs to adapt to the high concurrency, spike, and big promotion scenarios in the Internet scenario.

"Cloud-native middleware builds a new e-commerce platform for Amway based on the middle-office architecture of Internet business, gives full play to the technical advantages of cloud-native products, and supports a peak order value of 10,000 orders per second. Cloud-native middleware reconstructs the IT foundation of Amway's social business. , creating a benchmark for Amway’s global digitalization.”

Focus Media

Traditional servers cannot meet the new demands brought about by the rapid growth of Focus Media's business. They take too long, have low resource utilization, complex operation and maintenance, and require high personnel skills.

Time -consuming: The previous manual publication could not know whether the publication was correct or wrong in time, and it took a lot of time to check and revise;

Low resource utilization: The first publications are concentrated on Saturday and Sunday, so all resources are basically used on Saturdays and Sundays, and server resources do not need to be used most of the time, which leads to low resource utilization;

Complex operation and maintenance and high requirements for personnel skills: Due to the complexity of the business, the skills of relevant business personnel are also high, and more senior personnel need to be recruited to support the corresponding operation and maintenance work.

"For us, there are two options for going to the cloud. The first is to use K8s to build a container cluster by ourselves, and the second is to use function computing FC. If we use K8s to request cloud hosts, we need to build K8s ourselves, through external API to provide requests; with the serverless computing platform, we don’t need to care about how many servers or how many manpower are used, we only need to care about whether each API request arrives correctly, we can confirm whether the image is accurately recognized each time, and identify the image. The wrong thing is sent out and notified to the publication staff.

Alibaba Cloud Function Compute FC supports expansion to 7000+ instances within one minute. If we deploy K8s by ourselves, it will involve a lot of manpower and material resources, so we finally chose FC.

Automatic elastic shrinkage: You only need to set a processing volume of 2 million every Saturday and Sunday, which should be completed in two days. The peak is 9:00-10:00 am or 3:00-4:00 pm, and the automatic elastic shrinking of resources can be realized. ;

Resource-free operation and maintenance: no need for professional operation and maintenance personnel;

It can provide large-scale recognition capabilities: when we request the daily publication staff to publish at 6:00, 7:00, and 8:00 in the morning, computing power can be provided in real time.

In the future, we will also consider combining Serverless and Kafka for big data processing, which will be more efficient; we are also trying to use Serverless to solve the part where live video streams are pushed to video terminals in real time. "

pumpkin movie

A hit movie accelerated Pumpkin Movie's thinking about technological upgrading. After a movie was released, new registered users exploded, and the number of new registered users per day exceeded 800,000. The main traffic entrance, API, and gateway could not support it, followed by back-end services, databases, and emergency expansion of the entire link: the business recovered quickly, but the entire operation and maintenance process took 4 hours.

"At that time, there were two solutions in front of us. One was to build K8s by ourselves. Although it can solve the problem of high-density deployment, the learning cost of K8s is too high. It is easy to set up an environment to run and run. For production, it is necessary to form a professional team, which obviously cannot be completed in the short term. The second is the serverless application engine SAE. At that time, I felt that SAE did not need to be transformed, WAR/JAR package deployment, automatic elasticity, no need to buy machines, no need to operate and maintain machines, and to monitor security.

It took us three days from knowing about SAE, to communicating with Alibaba Cloud, and going online. On the fifth day, the deployment was successfully completed. By the seventh day, the remaining 30+ systems were quickly migrated to SAE in the same way.

Completed the Pumpkin Movie Serverless transformation in 7 days: In terms of elasticity, it will be automatically adjusted according to the user's optimization. The second is free operation and maintenance. The operation and maintenance speed of SAE is faster than that of manual labor. Finally, the release is faster and the monitoring is more complete. After using SAE, the operation and maintenance efficiency is increased by 70%, the cost is reduced by more than 40%, and the capacity expansion efficiency is increased by more than 10 times, which is an intuitive change brought to us”

02

Cloud-native technology has greatly lowered the threshold for digitalization,

Enables companies to focus on the business itself,

Without spending too much effort on the IaaS and PaaS level.

As basic cloud services have entered a mature stage,

Various upper-layer applications use cloud native as the technical base,

Gradually build an ecological closed loop of IT services on the cloud.

The emergence of cloud native is not accidental.

It is the only way for enterprises to digitally transform and upgrade.

And determine the outcome of the digital transformation of the enterprise.

From technology upgrade to scene landing,

The era of cloud native is coming.

"Cloud Native Practical Case Collection" is released!

For enterprises, cloud native is the shortest path to enterprise digital innovation. The value of cloud native for enterprise technology evolution is that it can realize cloudification of infrastructure, Internet-based core technologies, modern application architecture, and business intelligence. The most intuitive business value these features bring to enterprises are resource elasticity, system stability, application agility, business intelligence, and trustworthy security! Based on this, Alibaba Cloud released the "Cloud Native Practical Case Collection" to help more enterprises practice and implement cloud native through technology outlook, industry analysis, and interpretation of practical cases.

Click here to download the Cloud Native Practical Case Collection.