Three directions of cloud application architecture evolution

1、 Faced with opportunities and challenges, technology driven business innovation

Today, with the development of the times, enterprises in various industries are facing different opportunities and challenges.

The first concern is the comprehensive digitization of social lifestyles and production methods. Whether traditional enterprises or internet enterprises, their production systems, office systems, commercial sales, and customer interactions will be irreversible and fully online. Compared to today's delivery platforms or transportation systems, they can be operated by mobile phones to achieve online transactions.

Secondly, the external environment faced by enterprises is rapidly changing. For example, the preferences and needs of consumers are constantly changing with the level of consumption and the overall environment, and many retail enterprises also need to accelerate product launch and improve their core competitiveness, as is the case with To C's internet companies. At the Yunqi Conference in October, Alibaba Cloud's customers shared with Yingke that their new applications were launched as frequently as once a week.

Even without considering consumers, competitors are constantly competing for speed with each other. The current situation is also accompanied by uncontrollable impact of the epidemic, as well as regulatory policies and geopolitical factors that constantly affect the industry environment.

New technologies such as artificial intelligence, 5G, and big data have also given enterprises more capabilities and tools to innovate, achieve self transformation, and develop.

The comprehensive digitization of enterprises means that the key business and even the "life" of the enterprise have all been placed on the IT system. Therefore, enterprises need a stable and reliable system, a rapidly changing market, and a fierce competitive environment. They also need more agile research and development efficiency and application architecture to support business innovation, in order to win the lead in the market. It is even more necessary to make good use of new technologies such as AI and big data, optimize business management and operations, and achieve intelligent business upgrades.

All of the above require an excellent IT architecture to support. An excellent architecture should possess three characteristics: reliability, agility, and intelligence.



How can the cloud help enterprises build excellent IT architecture?

Each architecture pattern has its own suitable scenario, and the different stages of the enterprise, the number of human resources, and the skill level of developers will all affect the choice of architecture. There is no best architecture, only the most suitable architecture. And the architecture needs to continue to evolve in order to support business development.

Whether from the perspective of business requirement scenarios or technological trends, the overall development requirements of architecture are to be more reliable, agile, and intelligent.

More and more enterprises are attaching importance to the construction of high availability architectures, using rich methods such as dual activity, multi availability zones, multi regions, and chaotic engineering to improve the reliability of applications. Microservices and serverless have also been very popular topics in recent years. The above are all more agile IT architectures compared to before, and to some extent, they can also be called more reliable architectures. At the same time, the arrival of mobile internet and the Internet of Things has also led to a data explosion, and the demand for big computing scenarios such as big data and AI is also increasing.

So how can cloud computing help customers build a reliable, agile, and intelligent architecture?

01 Reliable

At the level of reliability, it can be divided into two parts: the reliability of the infrastructure layer and the reliability of the application layer.

The phrase 'weak foundation and unstable terrain' is widely circulated among architects. Alibaba Cloud provides very reliable basic resources and is the first cloud vendor in the world to provide a single instance availability SLA of 99.975%, and multi zone and multi instance availability of 99.995%. All of the above are thanks to Alibaba Cloud's self-developed Feitian base, which includes a large number of technological innovations, such as intelligent fault prediction with an accuracy rate of over 70%, and thermal migration technology with over 95% customer insensitivity.

The self-developed Dragon architecture by Alibaba Cloud also ensures a highly linear and stable performance of the entire system, allowing customers to experience predictable high performance while using Alibaba Cloud products.

Of course, the availability of single instance and single availability zone can only ensure high reliability in the current region, but it cannot withstand regional failures caused by extreme weather, fiber optic cuts caused by construction sites, etc. Therefore, customers also need to ensure high availability of the application layer and use disaster recovery solutions such as multi availability zones and multi regional deployments.

We suggest using mature managed products instead of self built ones, such as database and middleware products. Alibaba Cloud has invested heavily in these products, which are usually more reliable and convenient than self built ones. The ultimate high availability application must be a comprehensive architecture design, consisting of highly reliable basic resources, highly stable managed products, and the high availability design of the application itself.

Infrastructure not only needs to be highly reliable, but also needs to be transparent and open. After many customers go to the cloud, they feel that the infrastructure layer has become a black box, so they require the provider to clearly tell them what is happening in the underlying infrastructure, so that they can do better proactive operation and maintenance.

In fact, this requirement is very reasonable. Therefore, ECS will open up as much information as possible to users, encapsulating it into different interfaces and events, and providing them with the latest information on infrastructure such as cloud servers and operating systems at any time. The system predicts that the client's machine may go down, and detects that the CPU and memory have reached the warning line. Events will be sent, and the client can choose to subscribe.

According to user feedback, the most attractive aspect that attracts them to use Alibaba Cloud products for a long time is that Alibaba Cloud has a very rich and comprehensive interface. Alibaba Cloud's interfaces are by far the most comprehensive, rich, and detailed in China, and even globally, they are no less impressive.

02 Agility

What should we do if the world changes too quickly? The so-called world martial arts, only fast and unbreakable! The only way to face change is to be faster than change.

This requires an extremely agile architecture. Similarly, enterprises need to achieve agility at two levels: application level agility and infrastructure level agility.

Enterprises are building agile and flexible organizational structures and software architecture. One of the core ideas is to make R&D teams less dependent and easier to evolve independently and iteratively. Therefore, enterprises are beginning to adopt microservice architecture, which is also an application architecture in line with the cloud native trend.

In fact, the agility of application architecture also relies on the agility of infrastructure. The more agile the architecture, the faster it can respond to unusual business peaks, and the smoother the processing. This not only reduces costs, but also provides the best customer experience. For example, 12306, such as Weibo hotspots, this instantaneous level of traffic, even in offline data centers, even if the upper architecture is agile, it is difficult to deliver and deploy such a large amount of computing power in a short time.

Agile infrastructure requires two things: agile delivery and efficient management.

As a leading infrastructure, Alibaba Cloud must first achieve rapid delivery.

Alibaba Cloud elastic computing provides a variety of out of the box basic resources, with over a hundred specifications of cloud servers alone and providing ultimate flexibility. In July of this year, Alibaba Cloud, as the first and only cloud vendor to pass the large-scale cloud platform performance test of the China Academy of Information and Communications Technology, expanded its capacity to 10000 cloud servers in 18 minutes under the witness of the staff of the China Academy of Information and Communications Technology, which is not Alibaba Cloud's fastest speed yet.

At the Yunqi Conference in October 2021, research and development classmates of Alibaba Cloud's elastic container instance ECI demonstrated on site that they had expanded to 3000 PODs within 6 seconds. With the powerful elastic scalability of Alibaba Cloud's elastic computing, customers can quickly deliver and deploy underlying resources, easily cope with peak traffic or expand new business.

Alibaba Cloud also provides a variety of payment modes for different resource delivery methods, allowing customers to balance flexibility and affordability.

To truly achieve agility, customers need to be very efficient in managing and using computing resources, which requires various automation capabilities to support. From migration, deployment, operation and maintenance to capacity management, Alibaba Cloud provides a complete set of automated operation and maintenance tools that cover the entire lifecycle of resources.

For example, using Alibaba Cloud's resource orchestration service ROS, tens of thousands of cloud servers can be automatically deployed. During the epidemic, DingTalk used ROS to deploy over 10000 cloud servers in just 2 hours, smoothly overcoming traffic peaks.

03 Intelligence

Comprehensive intelligence includes intelligence at the business application layer and infrastructure layer.

At the business level, enterprises need to introduce corresponding machine learning, big data and other related technologies based on their own business, to achieve intelligent customer service, autonomous driving and other capabilities, which require a large amount of data and computing power as the foundation. To this end, Alibaba Cloud elastic computing provides customized big data and local disk instances for these scenarios, as well as GPU and NPU instances, providing the most suitable infrastructure for upper level business innovation.

At the PaaS level, Alibaba Cloud provides rich artificial intelligence services, machine learning, and big data frameworks, allowing customers to easily build upper level application intelligence.

At the infrastructure level, Alibaba Cloud's scheduling system, fault prediction, and operation and maintenance systems all widely use artificial intelligence technology, making Alibaba Cloud the world's leading IaaS technology platform. At the same time, in terms of user experience for elastic computing services, we also utilize artificial intelligence technology to provide customers with a smarter infrastructure.

For example, intelligent elastic services, in the past, customers may need to open their machines in advance for upper level business use based on their own business characteristics, or use Alibaba Cloud's elastic scaling services to set up strategies and schedule them to start.

The latest prediction mode of Alibaba Cloud's elastic scaling products can be modeled based on the CPU usage and internal network traffic of the user scaling group in the past 1-14 days. Machine learning algorithms are used to predict the overall usage in the next 2 days and automatically perform scaling operations. By using this feature, regular expansion and contraction customers basically don't have to worry.

Secondly, Alibaba Cloud's intelligent diagnosis and self-healing tools enable users to solve ECS related problems by submitting work orders or contacting service classmates, which takes a long time and to some extent affects the user experience. The instance health diagnosis tool, utilizing AI capabilities such as NLP on the backend, can help users quickly locate potential issues inside and outside the ECS and provide repair solutions, shortening the problem-solving cycle from 24 hours to minutes. In the field of cloud servers, Alibaba Cloud is the first cloud vendor to open up this full coverage diagnostic capability to users.

For example, Alibaba Cloud's intelligent resource optimization service can identify resource mismatches for users based on their resource usage, and recommend suitable instances for users based on business load. If the long-term low utilization rate of customer resources leads to cost waste; If the CPU load continues to be high, it may cause business instability, and it will recommend you to upgrade the configuration or add new resources.

Overall, Alibaba Cloud elastic computing is no longer just a platform that provides computing resources, but has evolved into a cloud platform that supports full application lifecycle services. Alibaba Cloud helps customers build a reliable, agile, and intelligent cloud architecture through its powerful and reliable cloud servers that cover all scenarios, efficient and intelligent automated operation and maintenance kits, and flexible resource supply. This year, Alibaba Cloud has also launched a shadowless cloud computer for office scenarios, as well as a cloud computing nest platform for partner services.

Using the cloud well to build an excellent application architecture can bring various value to the business, including helping the business to run continuously and smoothly, reducing business risks; Reduce costs; increase of efficiency; It also positively enhances the team's effectiveness and happiness.

Cai Junjie also shared two practical cases at the end:

Originally, Shentong Express used an offline computer room as a computing and data storage platform, and the demand for resources would expand every Double 11. After a major promotion, it would be idle and wasted. After going to the cloud, almost all resources are purchased according to quantity, and after steadily passing the Double 11, they are immediately released, truly achieving out of the box use without wasting a day. Compared with the traditional IDC architecture scheme on the same day of the 2020 Double 11 and 2019 Double 11, the use of the cloud based Dragon bare metal server+container service scheme helped Shentong Express reduce IT investment by 30% despite a significant increase in business volume.

Geely's automobile manufacturing enterprise has increased efficiency by 20% on the same hardware basis by using the Dragon Supercomputing Cluster SCC, reduced task queuing time by about three times, and the cluster size has expanded with business elasticity, ultimately significantly improving simulation efficiency by nearly 30%, shortening vehicle design and market launch time by several months.

2、 From cloud deployment to making good use of the cloud, seize the technological dividends

Going to the cloud has become a consensus in the industry.

Although cloud computing has been developing for over a decade, it is only the beginning. We have observed that many customers have not fully utilized the dividends and advantages of cloud, such as in the operation and maintenance field where cloud has changed the most, and most customers are still in the stage of semi manual and semi automated. So, now many enterprises have shifted their focus from using the cloud to making good use of it. We believe that in the next decade, making good use of the cloud will unleash huge technological dividends for enterprises.