1. Technical challenges
The emergence of the cloud and big data has given rise to a new generation of the information revolution. The emergence of distributed architecture has completely overturned the traditional concept of IT development for decades, and the informationized development of enterprises has become increasingly dependent on a distributed architecture. The distributed architecture uses the network to coalesce more storage resources and computing resources together and provides stronger support for the development of cloud and big data technology with the idea of "concentrating on doing great things". However, the biggest challenge lies in the network. Issues such as how to achieve efficient and reliable transmission of data in a distributed network environment and how to achieve intelligent collaboration of software and hardware resources to work efficiently in a distributed network environment has become the biggest challenges faced by distributed architecture at this stage.
2. Core value
The Bitsflow control system, based on centerless peer-to-peer network architecture, was completely independently developed by YOYOSYS. It was selected as one of the core technological breakthroughs in China's network information field in the "Five Years of Progress" large-scale achievement exhibition held by the 19th National Congress of the Communist Party of China. It supports the underlying network data transmission and wide-area distributed collaboration among various large-scale distributed application system components. It can ensure high performance and reliability of data transmission in complex network environments and support complex event driving, processing, and collaboration among software modules in distributed environments. Based on the centerless architecture design and logical address mechanism, the data control system is highly scalable. It provides complex network communication capabilities such as supporting group communication with strong consistency.
3. Technological innovation
a. Logical address to realize virtual network by building a 256-bit network addressing capability far larger than TCP/IP and mapping of the upper layer application to the underlying logical address, so that network communication can override TCP/IP and other various network protocols without impact on the upper layer business when the underlying network changes or fails.
b. Intelligent networking with the self-organizing ability and self-discovery and self-adaptation of multiple links. Self-organizing allows any member to automatically integrate into the network and normally work without other redundant configurations after joining the network. Self-discovery and self-adaptation of multiple links mean that it can automatically find a new unblocked path to ensure reliable data transmission when a certain network breaks down. Meanwhile, the optimal transmission path between two points in the current network environment can be automatically calculated to ensure the high efficiency of network utilization.
c. Virtual clock, which achieves absolute sequential network access to messages. Without dependence on any physical clock, the virtual clock is the group's logical clock so that the group members have a consistent concept of time. Communication messages are sorted according to the virtual clock to ensure absolute sequential data transmission.
d. Dynamic election to realize completely decentralized deployment of the entire environment. When most nodes in the group are destroyed, the rest of the remaining nodes can quickly regroup into a network that can work properly and continue to provide services to the outside world to ensure that the group's operation is not affected.
e. Topology reconfiguration to achieve a significant increase in consensus efficiency reached through multi-member interaction on the network. By reconfiguring the network topology, the complexity of forming consensus among members based on communication is significantly reduced. The traditional collaborative consensus that supports a maximum of tens of units in a controlled time is enhanced to a near-real-time collaborative consensus among tens or even hundreds of thousands of units.
f. Robust protocols, which overlay robust communication protocols on top of unreliable communication links to converting them into reliable communication links to achieve reliable data transmission. For example, superimposing robust protocols on top of unreliable network protocols such as MQTT protocols for IoT and UDP protocols for the Internet turns communication links into communication links with reliable protocols. Reliable data transmission can be ensured at the same time.
4. Application Scenarios
It has been used in many national ministries and commissions, public security organs, financial institutions, communication operators, and critical information infrastructures such as Beidou navigation system and power grids, deploying more than 1,000 data centers, more than 100,000 server nodes, and more than one million terminal devices, supporting thousands of business systems and carrying data volumes of more than EB level. It has become an important support for plenty of national critical information infrastructure such as the Beidou navigation system and power grids and a key support product to ensure reliable and smooth network operation and wide-area distributed cooperation in the current complex network environment.
Trailer of World Internet Conference
Copyright © World Internet Conference. All rights Reserved
Presented by China Daily. 京ICP备13028878号-23