Architecture
CONA is a new architecture, but one whose operations can be grounded in current practice. Its design reflects our understanding of the strengths and limitations of the current Internet architecture.
1. Architectural Principles
The hourglass architecture is what makes the original Internet design elegant and powerful. It centers on a universal network layer (IP) implementing the minimal functionality necessary for global interconnectivity. This so-called “thin waist” has been a key enabler of the Internet’s explosive growth, by allowing lower and upper-layer technologies to innovate without unnecessary constraints. CONA keeps the same hourglass-shaped architecture.
CONA works to deliver a consistent user experience, irrespective of the locations of computing resources. Consequently, ultra-high bandwidth, ultra-low latency, massive connections, and multi-service transport are integral to the provision of a high-quality computing network. The following architectural principles guide our design of the CONA architecture.
Elasticity: The traffic characteristics of a computing network diverge from those of the Internet, primarily due to a more pronounced requirement for elastic bandwidth. For instance, in a meteorological computing scenario, the meteorological center necessitates performing computations once or twice daily, with each computation lasting approximately two hours. Given the high bandwidth demand of these computations, the meteorological center would benefit from an elastic connection service offering adjustable bandwidth and customizable duration.
Agility: The widespread distribution of computing power necessitates that the computing network possesses the agility to access ubiquitous computing power. When enterprise and individual users access the computing network for computing services, they need not be concerned with the specifics of computing resources and their distribution on the network. Instead, their primary concern should be the ability to procure computing resources in an agile manner.
Lossless: The interconnection of computing power through the network is a critical process. Any packet lost on the network, or even within the distributed computing process of the cloud data center, diminishes the efficiency of computation. It is estimated that a packet loss rate of merely 0.1% could result in a 50% reduction in computing power. Therefore, lossless transmission within and between computing nodes is a crucial aspect of a computing network.
Security: Data, being the core element of computing, is a valuable asset that necessitates secure transmission to computing nodes, and the secure return of the computed result. Security is thus another key enabling factor for computing networks across various industries. This includes secure data storage, secure data encryption, secure data isolation between computing tenants, protection against external attacks and data leakage, and secure terminal access.
Authenticity: users would like to know that the computing resources and data came from the appropriate source, rather than from some spoofing adversary. Today this requires a PKI to provide users with the public key of the provider. Moreover, new authentification technologies like Zero-knowledge Proof (ZKP) can be adopted to secure the channel to the source, and automatically authenticate the data.
Openness: One of the key factors in the success of the internet is the allowance for users’ permissionless participation and the network's ability to adaptively expand without centralized control. Although not a relevant factor in the original Internet design, global deployment has taught us that “architecture is not neutral.” CONA makes a conscious effort to facilitate participants, empower end users, and enable competition.
2. The CONA Architecture
CONA is not simply a network that connects all computing nodes. It enables the computing power of all computing nodes to be aggregated into a computing power pool, thereby implementing global access and instant availability.
The components involved in the CONA include computing and the network itself. New factors such as 5G/6G, Network Functions Virtualization (NFV), Cloud-Native Computing and Blockchain are taken into account. And this is where the three major functions of the computing network come from.
Computing routes: The network can sense the computing power and provide the optimal routes for computing.
Computing scheduling: The computing network brain intelligently orchestrates and elastically schedules computing power resources on the entire network.
Computing transaction: a blockchain-based trusted computing power and network transaction platform.
Designing a distributed computing system is not only a computing problem, but is more of a network problem. This problem has two basic dimensions: computing and network. One should carefully consider the convergence of the two dimesions. Figure 2 shows the architecture of CONA following the layered principle of the internet.
Infrastructure Layer: produces computing power.
Convergence Layer: perceives and interconnects network and computing power.
Orchestration Layer: verifies, collaborates, orchestrates, and bills the computing power.
Service Layer: provides interface methods to enable users to access computing resources.
The technical composition of CONA primarily includes collaborative scheduling of the computing network on the control plane, network fusion perception on the data plane, and orchestration of computing resources on the management and service plane. The overall architecture of the computing power network system should possess the capacity to uniformly manage underlying computing resources, storage resources, and network resources. It should also be capable of measuring underlying infrastructure resources according to a unified standard, abstracting them as information elements loaded in network packets for sharing through the network.
In the current CONA system, consideration should also be given to providing intuitive components and service capabilities to users. This can be achieved by bridging the service layer with the underlying resources and network interfaces, thereby realizing visualization in verification, authentication, orchestration, and billing.
Last updated
