Computing-Network Integration

What Is Computing-Network Integration?

Computing-network integration seamlessly connects central and edge computing facilities via agile, reliable, and secure network connections. This approach ensures unified orchestration and control of multi-layer computing resources, facilitating fast, secure, and intelligent transmission of computing power across the network.

Multi-Layer Convergence Empowered by Computing-Network Integration

Computing-network integration signifies the fusion of computing and network resources, akin to the concept of Computing First Network (CFN). CFN, essentially a transmission network comprised of computing and network facilities, embodies the essence of computing-network integration. Architecturally, computing-network integration advocates a collaborative system characterized by four layers and six convergences.

Four Layers:

• Infrastructure: Utilizes computing and network facilities to perceive, connect, and coordinate diverse forms of computing power before integrating network and computing resources.

• Platform: An intelligent O&M platform integrates services and computing resources, facilitating rapid computing power scheduling.

• Application: Offers ubiquitous computing connections tailored for various vertical industries.

• Security: Incorporates endogenous security architecture design to ensure secure and reliable computing power scheduling.

Six Convergences:

• Service Convergence: The O&M platform integrates cloud-computing-network-security, guaranteeing computing service quality and real-time reporting of quality issues and computing resource distribution.

• Capability Convergence: Integrates various capabilities like intent awareness and fault monitoring to address diverse computing requirements in real-time.

• O&M Convergence: Consolidates network, computing, and storage resources into a unified pool, enabling a converged and intelligent O&M system.

• Data Convergence: Stores and manages various data types in a data pool, ensuring security and agile scheduling through big data learning and intelligent analysis.

• Computing Power Convergence: Provides computing power management, application, and visualization capabilities for heterogeneous computing entities, ensuring flexible application of ubiquitous computing power.

• Network Convergence: Ensures comprehensive computing-network integration through cloud-network-edge-device integration.

Why Is Computing-Network Integration Necessary?

The emergence of new digital infrastructure drives the transformation of the real economy. In the digital era, smart applications like biometric recognition, healthcare, and manufacturing are on the rise. This digital shift results in vast amounts of application data and increased computing demands. Projections suggest that by 2030, demands for computing power in unmanned driving, blockchain, IoT, and AR/VR will skyrocket, reaching 300 times the 2018 levels. Real-time and efficient computing is poised to define intelligent scenarios. Recognizing its significance, major countries worldwide are prioritizing efficient computing. By 2020, global computing power had surged to 429 exa-FLOPS (EFLOPs), a 39% year-on-year increase, positioning China alongside the US, Europe, and Japan.

Yet, challenges persist in computing facility construction, including high energy consumption, underutilization of resources, and uneven regional development. The digitalization drive exacerbates the disparity between computing supply and demand. This imbalance stems from two factors: a historical focus on individual computing advancements over connectivity, leading to fragmented end-to-end services, and an inadequate response to escalating demands through mere network capacity expansion. This approach fails to meet ubiquitous computing needs, resulting in local surpluses and global shortages.

Adopting networks to globally coordinate and connect computing power catalyzes network evolution towards intelligence and computing, fostering innovation in connection services. Thus, transitioning to network-based computing emerges as a pivotal direction for the telecommunications industry.

What Are the Characteristics of Computing-Network Integration?

Adaptive Computing Power Connectivity

In the future, computing power distribution will be pervasive, necessitating agile access via Computing First Networks (CFNs). SRv6 emerges as a key enabler, facilitating rapid access to computing power. It streamlines end-to-end path establishment across multi-tiered networks and expedites computing power transmission along low-latency routes. Moreover, SRv6 offers simplified access to computing power and cloud-edge-device resources with its one-hop-to-cloud and one-connection-to-multiple-cloud capabilities.

Reliable Computing Power Transmission

As computing power traverses networks, network faults may lead to packet loss, significantly impacting efficiency. To ensure seamless computing power transmission, lossless network transmission is imperative. Network slicing dedicates channels to guarantee lossless computing power transmission, providing:

• Implementation of Lossless Ethernet in hyper-converged data center networks to achieve zero packet loss.

• Secure isolation of slice tenants, ensuring reliable transmission of device-edge-cloud computing power within stringent timelines.

• Intelligent algorithms ensuring zero packet loss at full throughput, enabling 100% release of computing power.

Intelligent Computing-Network Orchestration

Upon computing power request initiation, the computing network brain evaluates network conditions, including bandwidth, latency, and reliability. If feasible, it employs multi-factor network measurement algorithms to compute real-time network paths based on SLA and bandwidth. This orchestrates unified computing and network resource allocation, monitoring computing power and network status for optimal transmission paths.

Adaptive Computing Resource Management

CFNs manage vast application connections, flexibly adjusting network quality for varied service scenarios to deliver high-quality computing services:

• IFIT deployment enables in-band service flow measurement and swift fault localization.字

• Utilization of APN IDs directs traffic to application-level slices or tunnels, ensuring differentiated SLA assurance.

Computing Security Assurance

A comprehensive security system, spanning network, computing, and security resource pools, detects and blocks threats in real-time, ensuring trustworthy computing power transmission:

• Leveraging devices, edges, and clouds to enforce secure computing resource access.

• Provisioning security resources during computing power scheduling, fortifying computing power transmission.

Dynamic Computing Resource Allocation

Agile network adjustments align computing resource supply with real-time user data volume changes, establishing elastic links between users and computing resources. Automated adjustments accommodate traffic fluctuations, reallocating surplus resources to high-traffic regions and restoring resources as traffic normalizes.