Why Networks Have Failed to Keep Up with Cloud Capabilities

As enterprises go digital, they are finding that their networks cannot keep up with computing demand.

By Zhang Shuai, TMTPost
Huawei Tech Issue 93

For as long as cloud has existed, cloud computing has hogged the limelight, while cloud networks have stayed distinctly low-key. But complex core services have imposed new requirements on cloud networks, so cloud-network integration is the only way forward if enterprises want to smooth out the bumpy road to digital transformation.

There are three key elements to cloud computing: computing power, storage, and networks. Computing and storage resources were virtualized long before networks, and so network technologies have lagged behind in the cloud era. 

Cloud computing divides and manages large-scale tasks through distributed computing. How is data transmitted and shared? How can computing power be distributed to where it's needed through networks? How can intelligence extend beyond data centers to cover enterprise production scenarios? The answers to these questions all rely on flexible and robust cloud networks.

As cloud computing matures, enterprise customers and vendors have turned to boosting network capabilities.

Misconception about digitalization: Prioritizing the cloud above networks

Cloud computing was introduced to the industry lexicon by Google in 2006. It was initially defined as computing resources that can be delivered over the network, like public utilities such as water, electricity, and coal. They are on-demand, pay-as-you-go services that put users at the center. As cloud vendors have focused on developing flexible computing resources, networks have been largely neglected and are failing to satisfy market demand. 

The cloud computing leader AWS is also considered an IT vendor. IT is key to the virtualization of computing and storage resources. That's why when cloud computing emerged as a product of ICT convergence, computing and storage started developing faster while networks were still barely able to meet demand. The last few decades have brought momentous changes, especially with the rise of Internet companies and cloud computing. However, telecom operators have remained the providers of dumb pipes. Today, enterprises want highly elastic cloud services, a demand that dumb pipes cannot meet.

One industrial enterprise in northwest China wanted to migrate to cloud as part of a move toward Industry 4.0. However, they could only accept a maximum of 5-ms latency, ruling out most cloud vendors, whose cloud data centers were more than 200 kilometers away from customers’ factories. Data centers have geographical limits, so if networks cannot support low latency across long distances, digital transformation is impossible.

Currently, most cloud vendors provide cloud network services such as traditional private lines and virtual private networks (VPNs), which are often criticized for poor network quality, lack of guaranteed latency and bandwidth, high costs, and rigid link provisioning and control. The inflexibility of cloud services caused by the network undermines the basic value proposition of cloud computing.

Interconnection of infrastructure does create value, but it’s more valuable to transfer cloud computing capabilities to enterprises that are undergoing digital transformation. This can be achieved by providing networks as a service and creating synergies between cloud, network, and edge.

Cloud-network integration is driven by the convergence of cloud computing and networks. When data moves to and from the cloud freely, it's up to enterprises to decide whether their data is computed in the cloud or at the edge. 

Whether computing is used to solve communication challenges or vice versa, cloud-network convergence is the way forward. Thus cloud vendors are shifting to cloud-network integration, which is both a practical need and a direct result of technological innovation. 

The foundation of digitalization: Intelligent cloud networks

Cloud-network convergence will pave the way for smoother digitalization journeys.

Four new challenges face networks in the cloud era: (1) The cloud is fast, but networks are slow. (2) It's hard to guarantee consistent experiences. (3) O&M is difficult. (4) Ensuring security is a challenge.

If an enterprise wants to migrate to a hybrid cloud or multiple clouds but the provisioning of private lines takes weeks or months, then it’s extremely hard to meet that enterprise's requirements. And if it wants to migrate its core production systems to cloud, SLA-defined network requirements are likely to be higher. The SLA needs to provide end-to-end cloud network assurance to ensure consistent user experiences, while providing intelligent network O&M and critical security protection. A small change at the network layer could have far-reaching effects.

When computing power is fed into networks across every industry, it needs to be as readily available as electricity. The first step to making this happen is meeting enterprises' requirements for digital transformation.

Networks will be digitalized, intelligent, and provided as a service. These are the three defining features of intelligent cloud networks, which are set to greatly boost enterprise productivity and provide new momentum to the digital economy.

Network digitalization: The entire network will be sensed, abstracted, and modeled in the digital world. Data on network status will be sent to cloud, so the entire network can be centrally managed in the cloud, and network status can be visualized in real time.

Network intelligence: When networks are digitalized, new technologies like AI and big data can infuse intelligence into cloud networks, enabling the intelligent, balanced scheduling of network resources and cloud resources. Intelligent network O&M can result in faster troubleshooting and more intelligent security defense, bringing network security protection to a new level.

Network as a service: This feature allows users to subscribe to network services in just one click and improves network response and cloud agility. Networks are made open for programming, so that they can connect with cloud services more flexibly to meet service requirements. Collaboration between cloud, network, and security will lead to secure cloud network services.

Cloud networks can connect with and enable cloud. When the cloud and network are properly integrated, the network can provide the resources required by the cloud and the cloud can invoke the network resources required by applications. 

Huawei has launched Intelligent Twins, a reference architecture for the intelligent upgrade of industries that aims to build one cloud for one city and one network for computing power. A single network can aggregate existing physical networks, such as IoT, government, and transportation networks, while logically isolating different types of data, like urban governance data, industry data, and population data. As such, Intelligent Twins architecture can lay the foundation for smart cities.

When helping customers in the financial industry go digital, Huawei advocates upgrading from traditional private lines to flat, high-quality private networks. This entails distributed architecture reconstruction, centralized resource management, and platform-based service capabilities, which has already enabled agile innovation in financial services.

Cloud networks as infrastructure

Intelligent cloud networks are about to take their place as part of the critical infrastructure of the digital age – all vendors in the industry hope to carve out a share of this growing market. The major players in this market fall into three categories: telecom operators, Internet companies, and digital transformation service providers.

Telecom operators build a cloud WAN for tenants through cloud backbone networks deployed in cities with network virtualization. Operators can provide high-speed interconnections between clouds, connect SD-WAN branches and private lines to the cloud, and offer other products and services. They enjoy unique advantages in networks.

Globally, Internet companies are first movers in cloud computing and have attracted the first batch of customers to cloud. However, as the cloud computing industry moves into a new phase of development, Internet companies cannot just copy their successful experience in the Internet industry over to other industries. Traditional industries have their own barriers to cloud migration, requiring Internet enterprises to upgrade their service capabilities if they are to grow.

The third type of player to emerge as industries go digital is the digital transformation service provider. These companies, such as Huawei, understand both the cloud and networks. According to Gartner, HUAWEI CLOUD was the fastest growing major cloud service provider in 2020, ranking second in China and in the top 5 in the global public cloud IaaS market. To date, HUAWEI CLOUD has launched more than 220 cloud services and 210 solutions.

Huawei is able to leverage the network capabilities it’s built up over decades, helping bridge the gaps in cloud network capabilities and improving the underlying infrastructure of cloud computing.

The true convergence of computing and networks will enable the digital transformation of all kinds of industries. More computing resources with available to cloud, enabling data migration between service scenarios and networks that can transmit computing power to different nodes. In turn, this will enable the true digital transformation of industries.