In 2001, 22 years ago, Feng Yong, a young engineering man, decided to join F5.
Feng Yong, Technical Director of F5 China Architect Department
Founded in 1996, F5 is a technology-based company focused on application delivery, headquartered in Seattle, Washington, USA, and entered China in 2000. The rapid development of the Internet at the beginning of the new century has brought great opportunities to F5. Feng Yong, who has worked in Wanda Information System Integration Department and other companies for several years, believes that F5 has a unique core technology of load balancing that occupies an important strategic position in the future IT system, and it is very important to the Chinese market, so it is worth joining.
Forge ahead in the rapids and continue to innovate. This has been done for 20 years. China's IT application market has grown rapidly from traditional data centers to the era of cloud computing with the rapid development of the Internet and mobile Internet, and Feng Yong has also evolved from a simple system integration master to a "multi-cloud trader" standing at the forefront of digital technology ": Served as the technical director of the architect department in F5 China, and was called the "Multi-Cloud God" by his colleagues and industry friends.
Cloudiness will inevitably become the norm and the main theme of the era of digital intelligence. However, in the eyes of the Great God, although cloudy has been talked about for many years, in China, the real cloudy war has just begun.
Its opening will bring huge changes to China's cloud computing market and even the entire digital intelligence ecosystem...
cloudy myth
What makes the master a little "worrisome" is that in our interviews and investigations, we found that multi-cloud has become a hot topic in the industry, but there are many misunderstandings about it.
Is multi-cloud popular in China?
Haibi Research Institute found a strange phenomenon when investigating the status of multi-cloud applications. According to the data released by many foreign research institutions such as IDC, more than 70% of users have begun to deploy multi-cloud, and at the same time, they are connected with more than 2 or 3 public cloud services. business cooperation.
However, the Haibi Research Institute’s survey data on domestic industry users shows that in domestic enterprises and institutions, the proportion of multi-cloud applications is less than 10%. On the one hand, there are many enterprises that have not migrated to the cloud and are still deploying applications in traditional data centers; on the other hand, among the enterprises that have migrated to the cloud, the proportion of using a cloud is still the vast majority.
However, it is worth noting that 75% of enterprises have begun to consider multi-cloud and have plans to deploy multi-cloud.
Is your unit really not cloudy?
This is also a highly controversial topic. The above multi-cloud application status data is the answer given by Haibi Research Institute through interviews with CIOs or relevant persons in charge of informatization and digitalization.
But in enterprises and institutions, there is another fact that many organizations have begun to use various public cloud-based SaaS services. For example, various instant messaging office applications such as WeChat and DingTalk, and various digital marketing services such as customer acquisition, advertising, and promotions that some sales departments are using. These SaaS services are also "clouds", but they may not be included in the "multi-cloud" category by CIOs, because they do not belong to infrastructure-level clouds such as IaaS.
But in essence, this is also the state where users are actually using "multi-cloud". There will be inextricable relationships between different SaaS, SaaS and the underlying technical infrastructure, including IaaS and data centers. According to this understanding, the proportion of multi-cloud applications may be much higher, even exceeding 80%. Because corporate office communications such as corporate WeChat and DingTalk have become very popular in all walks of life.
Are cloud islands cloudy?
Haiby Research Institute also found in the survey that there is another situation that is very common among existing "multi-cloud" users. They have used two or more clouds, but each cloud service provider is responsible for different business systems. There is no connection with each other, and these different clouds form a "cloud island", which is a bit like a "chimney system" in the information age.
For the entire organization, they are indeed "multi-cloud", but this "multi-cloud" does not give full play to their respective advantages and does not carry out corresponding resource integration. In specific applications and operations, it is still in the state of a cloud in essence.
In Dashen's view, these are not his ideal cloudy state. It cannot be simply understood that cooperation with multiple public cloud vendors is multi-cloud.
From the Internet to APP to the Internet of Things
It can be said that Dashen himself is the first batch of technicians in the Chinese market to contact public cloud services.
Accompanying F5 to continuously implement application delivery projects for customers, he has personally experienced Chinese enterprises from traditional data centers to private cloud services, to public cloud services, hybrid cloud services, and to today's multi-cloud services, IoT + AI edge cloud The whole process of service.
It is precisely because of these rich experiences that the master has a deep understanding of the core of the public cloud and the development of the public cloud. In his opinion, the reason why Duoyun has such misunderstandings is that it is closely related to the development of Duoyun in China.
The development of multi-cloud in China can be roughly summarized into the following four stages:
First , the transition from traditional data centers to private clouds. This process is a period of rapid development of virtualization vendors represented by VMWare and SDN vendors represented by Huawei. The application of virtualization technology has greatly improved the cycle and deployment speed of customer application development, and further simplified customer network management tasks, providing technical support for customers' transformation from traditional single application to B/S architecture.
In fact, F5 also seized the transformation opportunity of customer applications from minicomputers to PC servers to virtual machines during this cycle, and vigorously helped customers implement virtual machine cluster services to improve the reliability and scalability of web applications, thus becoming a A leader in application delivery for 10+ years.
Second , the transformation from private cloud to public cloud. This process is represented by the global public cloud development led by AWS/Azure/GCP and the domestic Alibaba Cloud, Tencent Cloud, and Telecom Cloud catching up. A major feature of this stage is the development of mobile apps. The rapid iteration of App applications and the requirements for user response time have led more and more customers to try to extend App from private cloud/data center to public cloud to improve deployment time and scalability faster.
Representative vendors at this stage include VMWare and IBM. VMWare provides services on multiple public cloud platforms through its own virtualization platform, thereby helping customers to smoothly migrate virtual machine-based application services between data centers and public clouds. Similarly, IBM/Oracle also provide public cloud database services for their private cloud/traditional customers in the form of public cloud.
A more classic domestic case is Taobao. At this stage, Taobao built its website on the Alibaba Cloud platform and successfully migrated from the traditional data center to the public cloud, thus providing a technical foundation for the later Double Eleven miracle.
During this period, F5, where Dashen works, cooperated with multiple public cloud vendors to provide unified application delivery and security services on almost all public cloud platforms, helping customers' application services to migrate smoothly and quickly to the public cloud.
Third , with the launch of AWS in China in 2016, enterprise customers began to choose multiple public cloud services based on technical characteristics and regional coverage. Enterprises will choose different cloud service providers according to the region where their customers are located, such as Alibaba Cloud and Tencent Cloud in China, and AWS/Azure/Google Cloud in overseas. At the same time, some customers will choose GCP or Azure's AI service and Akamai's CDN service based on their technical advantages.
During this period, the container-centric PaaS platform has been vigorously developed, providing technical support for the rapid and smooth migration of user application services to multi-cloud platforms, and also promoting the transformation of traditional applications to modern applications.
Fourth , after entering 2018, enterprise customers began to consider edge cloud, Internet of Things and cross-cloud SaaS services. At this stage, enterprise customers have begun to choose private cloud, public cloud, and edge cloud to provide services based on the characteristics of different application services.
According to F5 China CMO Jack Wang, F5 has also released its own SaaS distributed cloud service XC for these enterprises. Its purpose is to make full use of its technical advantages across multi-cloud service providers to help enterprises better implement unified deployment, delivery, and management of application services on multi-cloud.
Of course, the four stages of multi-cloud will be reflected in different ways for different customers, and even most customers will have multiple stages at the same time, which is also determined by the complexity and flexibility of multi-cloud application services.
This presents both a challenge and an opportunity for organizations across industries. By seizing this transformation opportunity, we can help enterprise customers achieve digital transformation, realize the transformation from traditional applications to modern applications, and fully enjoy the convenience of huge, cost-effective computing resources and flexible business expansion brought by multi-cloud.
At the heart of multi-cloud is distributed cloud
From the experience of the Great God, "multi-cloud" (multi-cloud) can have three levels:
The first level is that an organization is cooperating with multiple cloud service providers at the same time, but there is no connection between them, just a "cloud island";
The second level is that an organization is cooperating with multiple cloud service providers at the same time, but there is no in-depth contact with each other, and only some interfaces are made through APIs to achieve exchange or sharing at the data level;
The third level is that an organization is working with multiple cloud service providers at the same time, and they have in-depth cooperation with each other, and can use a "distributed architecture" to realize the connection and cooperation between different cloud services.
The third level is the ideal "multi-cloud" world, which can maximize the value of multi-cloud. Of course, every tier of multicloud has value. For example, the first level can realize the centralized procurement of cloud service provider resources, which directly helps to improve the service level of cloud service providers and reduce prices; the second level goes further and can realize data sharing among different clouds. It can improve data-driven capabilities and maximize the value of data; the third level has the greatest value. It can not only realize the first two values, but also make the most efficient use of the characteristics of multiple cloud service providers to improve the availability and value of resources. .
To some extent, distributed cloud is the technical basis and core of multi-cloud architecture, which can achieve high availability, high performance and high scalability of multi-cloud architecture, and can better realize the deployment and orchestration of modern applications. Without the support of distributed cloud, multi-cloud architecture cannot give full play to its advantages.
For example, in a multi-cloud architecture, enterprises may need to disperse applications and data across multiple cloud platforms to meet the needs of different businesses. In this case, without the technical support provided by the distributed cloud, it will be difficult for enterprises to achieve load balancing and high availability guarantees, and may face issues such as data consistency, security, and regulatory compliance. Distributed cloud can effectively help enterprises overcome these problems and improve the reliability and security of multi-cloud architecture.
Therefore, in the world of digital intelligence, if an organization's multi-cloud strategy does not have distributed cloud support, it is not an ideal multi-cloud architecture.
The difficulty of distributed cloud is multi-cloud orchestration
According to the investigation of Haibi Research Institute, there are currently four typical multi-cloud application scenarios:
One is the migration of traditional data centers to private clouds and container clouds, but not public clouds;
The second is that there is a private cloud but must be migrated to the public cloud;
The third is the migration of one public cloud to multiple public clouds, which is especially obvious when Chinese companies go overseas, and they need to use different public clouds in different regions;
The fourth is to develop towards the Internet of Things, adopt edge cloud, and realize collaborative services between different carriers. For example, when a car enters a gas station, some relevant information about refueling can be automatically downloaded on the mobile phone.
Once an organization has multiple public clouds and adopts a distributed cloud architecture, it will involve connecting these different clouds and performing unified orchestration and management of the resources they provide. In the era of distributed multi-cloud, the orchestration of business systems will be very different from the orchestration of traditional application systems, so it will face more challenges.
From the experience of the master, there are four aspects that are often encountered in the current multi-cloud deployment.
First, unified management of multiple clouds is difficult. Generally speaking, different clouds have different policies, different configurations, different traffic, and different APIs. It is a relatively complicated matter to manage them in a unified manner.
Second, unified traffic management is more difficult. Where is the peak traffic coming from? Which traffic is placed on which cloud? Does whoever is cheap use it? How to dynamically allocate resources? This requires the accumulation of experience and wisdom to do well.
Third, multi-cloud security strategies are more difficult. Different clouds often have different security policies. But for multi-cloud enterprises, different security policies are easy for hackers to find loopholes. Therefore, how to ensure the consistency of security policies is very important.
Fourth, it is a big challenge to the professional ability of technicians. Because the multi-cloud world requires technicians to master the technical challenges of multi-cloud, and to have an in-depth understanding of the technical architecture, advantages and disadvantages of each cloud. It is difficult for ordinary technicians to master all of them.
The multi-cloud battle is on
The rich scenarios and problems of multi-cloud are bringing huge markets and opportunities, attracting cloud players from all walks of life to show their talents. Dashen told China Software Network that he was in it and clearly felt the excitement of the multi-cloud market, where warlords from all walks of life were fighting fiercely.
When we imagine ourselves standing at a high altitude and overlooking the entire cloud service market, we can find two particularly distinct multi-cloud forces.
The first force is public cloud providers represented by GCP (Google Cloud) and Huawei Cloud. What these public clouds have in common is that they are latecomers compared with the pioneers of cloud services such as AWS in the United States and Alibaba Cloud in China. Moreover, their market share in the market is much lower than that of AWS and Alibaba Cloud. In order to allow their cloud services to penetrate into organizations that have been occupied by AWS and Alibaba Cloud, they vigorously promote the multi-cloud strategy to these customers.
Public cloud giants such as GCP and Huawei Cloud have launched very useful tools and solutions for multi-cloud strategies. China Software Network found that the most special performance is Anthos launched by GCP.
April 9, 2019, San Francisco, USA. The Google Cloud NEXT 2019 conference is officially open. This was the grand meeting of Google Cloud before the epidemic. It was also at that meeting that Pichai, head of Google, positioned "multi-cloud" as one of Google Cloud's most important differentiation strategies, and launched Anthos, a multi-cloud management platform based on Google Cloud services, at the conference.
HUAWEI CLOUD has also launched products and solutions such as a multi-cloud management platform and "WAN middle platform", which provide common management functions such as multi-cloud routing and security, and realize unified management and distribution of cloud resources in a multi-cloud environment.
The second force is multi-cloud service providers represented by F5, IBM Redhat, VMware, and Yunke Tongminghu. The common feature of these multi-cloud service providers is that they are not public cloud service providers themselves, but they provide organizations with a unified platform for multi-cloud orchestration and multi-cloud management.
The multi-cloud management platform is a just-needed product for organizations with multi-cloud deployment, and it will also generate a huge market. Compared with public cloud multi-cloud management platforms such as GCP and HUAWEI CLOUD, their products are more independent. Dashen believes that in the next five years, a giant company will definitely be born in this market.
At present, these companies have launched related products and solutions. For example, F5 has products such as BIG-IP and NGINX Controller, IBM Redhat has Openshift, VMware has vRealize Suite, etc.; and Yunke Tongming Lake, as a representative of domestic multi-cloud service providers, has also launched CloudOS. These multi-cloud management tools can help enterprises realize various management functions, including resource management, automated deployment, security management, cost optimization, etc.
Relatively speaking, the strengths of IBM Redhat and VMware lie in the hybrid management of private and public clouds, while the strengths of F5 and Yunke Tongming Lake lie in the management of multiple public clouds.
Interestingly, in the face of the irreversible trend of multi-cloud, China Software Network found that different cloud vendors also showed very different coping strategies. These are important factors that will affect the future development of multi-cloud and the competitive landscape.
For example, Jassy, the former CEO of the global cloud giant AWS, believes that multi-cloud is indeed a trend and will become a common choice for every organization. However, if an organization uses multiple clouds, it will not and should not adopt an equal distribution strategy. Instead, it must adopt a "one master and multiple configuration" strategy, that is, one public cloud is the main one, and other public cloud service providers are supplemented. It could be 73, 82 or even 91, but not 55.
Of course, for the choice of the main public cloud, he believes that the leader with the largest market share should be the top priority, that is, choose AWS.
Domestic cloud giants also follow the general trend, agree with the multi-cloud trend, and launch multi-cloud solutions. But for their own customers, the major giants are still trying to find ways to increase customers' dependence on their own cloud services through many value-added services.
Some companies once questioned that the current public cloud giants want to lock in customers as an initiative to win the multi-cloud market.
However, some people in the industry believe that as an organization in the era of digital intelligence, it is inevitable to be locked in by cloud service providers or other manufacturers, and it is very difficult not to be locked in. So, ideologically, every organization should accept being locked down.
As long as you use cloud services, you will either be locked in by cloud giants such as AWS and Alibaba Cloud, or locked in by professional technicians after adopting multi-cloud. Because multi-cloud management is too complicated, it must closely rely on skilled technical personnel, which means that the multi-cloud architecture is easily locked by technical personnel. Relatively speaking, the risk of being locked by cloud giants is lower than being locked by technicians.
In addition, multi-cloud also brings higher operation and maintenance costs. Therefore, a multi-cloud architecture is not necessarily the best choice.
From this perspective, multi-cloud is not inevitable. It is also destined that the multi-cloud market will be a fierce battlefield full of appeals and struggles from all parties. Multi-cloud forces, cloud pioneers, who will win the war in the future? Facing different forces, how should enterprises choose? how to respond?
Win in the cloudy trader
God did not give a prediction of who will win the war in the future. After all, the era of digital intelligence is changing too fast. But he believes that F5 will be the star of hope in the future.
How should organizations respond to a multi-cloud world? God gave his advice. He believes that in the face of a complex multi-cloud world, it is very difficult for an organization to achieve complete control. The world of digital intelligence is actually very fragile. Sometimes a small loophole can bring down an entire digital enterprise.
Therefore, Dashen suggested that when an organization deploys multi-cloud, it must first realize that there is a high probability of occasional problems. What enterprises need to do is to be prepared. In case of a problem with any cloud, how to deal with it and how to ensure business continuity? Have a contingency plan.
Second, it is necessary to strike a balance between cost and safety. The introduction of multi-cloud can increase the security and reliability of cloud services, but at the same time it will also increase the complexity and difficulty of management, and may also bring about a certain increase in costs in the short term. Therefore, organizations should pay attention to the timing and quantity of introducing multi-cloud, and strike a balance between cost and security.
A senior architect of the Agricultural Bank of China even told China Software Network that for many large central state-owned enterprises, "island cloud" may be the best multi-cloud model at present. You can enjoy the better prices and services brought by multi-cloud, but will not be affected by the technical complexity brought by multi-cloud.
Finally, and most importantly, you have to be a talented team of multi-cloud traders. Although it is very difficult to fully control the digital intelligence platform and digital intelligence technology, the master believes that it is still necessary to approach complete control as much as possible and infinitely. Only in this way can we truly deal with problems calmly when we encounter them.
Cloudy Trader is a new role. Before the multi-cloud era, the R&D department, operation and maintenance department, and security department were often separated, but now they must be together to implement SecDevOps. This requires multi-cloud operators to have more technical capabilities. For example, you must be familiar with cloud-native and modern application architecture; you must be familiar with CI, CD, and the framework of continuous development and continuous integration; you must be familiar with data security, privacy security technologies and products; you must be familiar with automated and intelligent AIOps operation and maintenance management platforms...
Dashen believes that it is more appropriate for the multi-cloud operator in an organization to be led by the CIO, and a position can also be set up under the CTO. Relatively speaking, CIO is the most suitable. But CIOs need to be more familiar with the business, which is also a major challenge for CIOs in the multi-cloud and digital era. Furthermore, a multi-cloud trader is not an individual, but a representative of a team. What CIOs need to do is to lead these multi-cloud operators to gallop in the turbulent digital world.
Facing the increasingly complex multi-cloud world and the increasingly powerful ChatGPT, the digital intelligence world is becoming more and more uncertain under the pavement of new technologies such as cloud computing and AI. For the operation of an enterprise or an organization, it is necessary to find relative certainty in the long-term uncertainty. How to get certainty? This requires multi-cloud operators to use their professional capabilities to find ways to find out amidst technological changes.
The Great God said that the Duyun battle curtain has just opened, and the Duyun traders are dancing. Whoever captures the heart of the cloudy trader will win in the cloudy battlefield.
Click "Read the original text" to enter the small survey of multi-cloud traders