With the advent of the digital economy, cloud computing is no longer seen as “old wine in a new bottle”, but the infrastructure of the future economy. IDC recently in netease cloud and the conference of the digital economy, innovation lead – 2018 China enterprise digital development report (hereinafter referred to as the “report”), points out four big features of the digital economy including “to the cloud as the core technology of the engine, the data as the key factors of production, ecology as the main commercial carrier, in order to open a win-win cooperation model” for the mainstream. Large and small cloud computing manufacturers have also offered their own digital blueprint, such as AWS “enabling digital transformation”, Ali Cloud positioning as “digital transformation expert”, Qingyun “Reinventing ICT to support enterprise digital transformation”, NetEase Cloud “assisting enterprise digital innovation”.
In this context, cloud computing platforms also have new development directions, most obviously including open cloud, full stack cloud, elastic cloud and hybrid cloud — except for full stack cloud, other directions are not new concepts, but also have new connotations, and many cloud computing vendors are stuck in these areas.
(NetEase Cloud and Inspur jointly released the first all-in-one cloud computing full-stack machine)
Open cloud — Open ecology wins
This may be a surprising proposition – isn’t cloud computing inherently open? But that is not necessarily the case. Li Jin, vice president of Ali Cloud, said, “One is the cloud of doctrine, one is the independent and controllable flying cloud”, which ignited the flames of open source and self-research. Ali Cloud claimed that autonomous and controllable (self-research) cloud can go further, but open source cloud computing schools have fought back, Huawei Cloud president Zheng Yelai wrote that only dare to open will be strong, can be strong, if the emphasis on autonomous and controllable do not use open source code, including MySQL. Even the low-key NetEase Cloud also joined the debate. Wang Yuan, vice president of NetEase and executive director of Hangzhou Research Institute, said that from the perspective of smooth evolution, (customers) independent control and ecological talent, “Enterprises must use cloud based on open source technology and meet the most mainstream open standards. So the future must belong to open cloud “. As a firm open line of evidence, NetEase cloud released based on open source, compatible open source light boat micro service, hanhai private cloud.
Sustainability is only one aspect. Based on the mainstream open standard interface, the open ecosystem required by the digital economy will have a solid foundation, and the complete solution required by users will be easy to integrate. After all, no software empire can cover every aspect of enterprise operation. Aliyun chose to develop its own software in 2009 because “there was no open source software on the market capable of addressing the scale, efficiency and security of human computing”. However, in hindsight, Open source cloud computing has become very mature. OpenStack, Ceph, Open vSwitch and Kubernetes are all technical frameworks supported by many big manufacturers. There is no need to duplicate the wheel. Therefore, Huawei, Tencent, China Mobile, NetEase and other companies have adopted OpenStack technology. Even Microsoft, which used to regard Linux as “cancer”, has its Azure platform and open source software inseparable. In fact, Ali Cloud has not completely rejected open source, its virtualization layer uses KVM, its container, database, cache layer also supports Kubernetes, MySQL, MongoDB, Redis.
Therefore, the trend of open source cloud and open cloud is irreversible. Cloud computing competition is ecological competition, closed system does not mean that there is no ecology, but relying on only one company, the momentum is not enough to open ecology. While this may not be obvious for service-heavy public clouds, for product-heavy private clouds, openness is even more important.
Full stack cloud — IaaS+PaaS is not enough
To meet the development needs of digital economy, the significance of cloud computing has gone beyond computing and storage. IDC report believes that digital innovation is the core of the digital economy, and the key technologies enabling digital innovation include the Internet of Things, cloud computing, big data/artificial intelligence and security. Cloud computing, as the foundation, provides programmable infrastructure, advanced processes and methods, and service-oriented software architecture. Improve the efficiency of office coordination, product research and development, production and transportation, sales service and other processes. In other words, for digital innovation, enterprises need full stack capabilities. For example, the industrial field hopes to software and tool the experiential knowledge through micro-services, and build a universal platform with the help of specialized tools to realize the reuse and promotion of industrial knowledge and drive the improvement of the overall level of the industry. IDC report also gives examples of wanxiang’s rapid development and deployment of industrial APPS through microservice architecture.
Forrester’s survey also shows that full-stack cloud platforms are increasingly favored by Industry users in China. Its full-stack cloud platform is defined as a cloud service capability that not only covers IaaS and PaaS, but also provides network communications, hybrid cloud solutions and full life cycle management, as well as the capability to support new applications such as artificial intelligence and Internet of Things.
Qingyun has raised the banner of enterprise full stack cloud ICT service provider. “The core of cloud is not ‘cloud for cloud’s sake’, but to re-create an ICT system in the form of cloud to support enterprise digital transformation. Today, Qingyun is positioning itself as an enterprise-level full-stack cloud ICT service provider.” Qingyun CEO Yunsong Huang said. In response, QingCloud released nine product brands, including QingCloud Platform, Qingcube Hyperconvergence System, KubeSphere Container Management platform and RadonDB distributed database, to show its ambition.
Taking the differentiated competition route of “scene-oriented cloud service”, NetEase Cloud has also entered the track of full stack cloud. NetEase Cloud focuses on “the most complete micro-service solution in the industry”, which integrates container platform, DevOps, AIOps, automated testing and other capabilities in addition to micro-service framework and API gateway. Its Hanhai private cloud is also on the IaaS, PaaS on the integration of light boat micro services. Not only that, NetEase cloud also jointly launched the industry’s first cloud computing full stack all-in-one machine — Hanhai All-in-one machine. Through the integration of Hanhai platform and infrastructure, the integration of soft and hard delivery is promoted to the delivery of architecture evolution and business research and development. Internet companies involved in hardware, its adherence to digital innovation and full stack cloud, can be seen.
In addition, Huawei also believes that “industry customers need full-stack cloud” and has released a full-stack private cloud solution. In Huawei’s view, the full stack cloud covers the full stack service bearer, service, resource management, architecture evolution and other capabilities. Only the full stack cloud can better support the whole process of enterprise users’ cloud services and provide continuous support from the perspective of architecture and resources.
Elastic Cloud — Applications need to be elastic
The definition of cloud computing cannot be defined without elasticity. A cloud without elasticity can be said to be a pseudo cloud. Traditional elasticity refers to the elastic use of cloud resources according to the volume of services, but the elasticity of full-stack cloud platform is different. The first is the level of elasticity. The elasticity of the full-stack cloud is not only Elastic Compute Service (ECS) and Elastic Block Store (EBS). However excellent the elasticity of computing and storage is, it only achieves the elasticity at the resource level. In the service-oriented architecture, enterprises need the elasticity at the Service level more. Or apply elasticity. For example, the elastic expansion of NetEase Cloud Hanhai all-in-one machine is not only the computing node, but also the capacity and quantity of services, which benefits from the maturity of container technology. Some cloud designs for container products are lighter virtual machines with abstract granularity, speed, and flexibility, which are indeed the advantages of containers over virtual machines and also in line with user habits, but do not meet the best practices of application elasticity.
Second is the scale of elastic scaling support. With the development of China’s Internet for more than 20 years, various ways of playing have become mature. Not to mention double 11, any action of big V may detonate business. For example, Lu Han’s love affair once let microblog mixed cloud “hit in the face”, which tests the scale support ability of cloud platform. At a very large scale, many of the concerns of an enterprise can be exposed, such as a recent cloud computing incident in which disk errors and irregularities during relocation and expansion resulted in the loss of customer data. Therefore, in terms of scale support, NetEase Cloud Hanhai private cloud supports a single cluster of 5000 physical nodes and a single tenant supports 30,000 cloud hosts. According to Wang, 5,000 nodes is now twice the size of the second largest node. On the other hand, its Kubernetes single cluster supports 30,000 nodes and 450,000 containers, and the upper limit of application layer size has also been broken.
Hybrid Cloud — Embracing a multi-cloud architecture
In the traditional hybrid cloud, the private cloud and a public cloud communicate with each other and manage resources in a unified manner. For example, Microsoft AzureStack supports only Microsoft Azure public cloud. However, in the digital economy environment, with the increasing importance of cloud computing and sudden public cloud failures, a single public cloud, even with multiple AZs (availability zones) or regions, cannot meet customers’ requirements for service continuity or data reliability. Therefore, cross-cloud backup has been discussed.
A multi-cloud strategy, for cloud hot migration or failover or avoiding vendor lock-in, is superior to a single cloud. Intel is given from the Angle of optimizing enterprise workloads, in addition, released its 2017 “in the public Cloud, hybrid and private Cloud reasonable placed between the workload,” white paper points out that every enterprise and industry needs and goals of each are not identical, real Cloud strategy applicable to all customers, it should be “cloudy” (Multi – Cloud) strategy, Choose the right cloud platform for your existing IT workloads based on their characteristics and needs before moving them to the cloud.
At the product level, we can see that qingyun launched nine major products, including an iFCloud unified multi-cloud management platform and an OpenPitrix multi-cloud application management platform. The former is aimed at multi-cloud environment, reduce the implementation threshold of enterprise cross-cloud services, achieve IT management cost reduction and efficiency; The latter is an open source platform for managing different types of applications in a multi-cloud environment, enabling efficient application delivery and operation management.
NetEase Cloud Hanhai private cloud also brings a new breakthrough — it can not only seamlessly access NetEase’s public cloud, but also support access to all mainstream public clouds, there will be open to the end. Of course, this is also due to the openness of OpenStack.
Huawei proposes that the full-stack cloud is based on a unified distributed architecture, which not only supports unified management from multiple data centers to edge cloud nodes, but also supports the long-term evolution of enterprise services by connecting public clouds through hybrid clouds.
It can be expected that there will be more and more cloud service providers and cloud products that support cross-cloud and multi-cloud services.
The above summarizes the four directions of cloud computing development worthy of attention under the digitalization wave from the perspective of platform. Digital economy itself includes two directions: digital industrialization and industrial digitalization. On the platform, the deep integration of cloud, big data, artificial intelligence, Internet of Things and the industry is also worth paying attention to, and such integration will also affect the choice and evolution of the platform. Therefore, in the process of platform selection, enterprises need to conduct repeated and comprehensive arguments according to the specific situation, which will not be described here.