P2P technical architecture and Classification

P2P technology has three architecture modes: centralized directory structure represented by Napster, pure P2P network structure represented by Gnutella, and hybrid P2P network structure. From the perspective of P2P technology, the P2P technology so far can be divided into four generations: the first generation P2P (Central Control Network Architecture), and the second generation P2P (divided

P2P technology has three architecture modes: centralized directory structure represented by Napster, pure P2P network structure represented by Gnutella, and hybrid P2P network structure. From the perspective of P2P technology, the P2P technology so far can be divided into four generations: the first generation P2P (Central Control Network Architecture), and the second generation P2P (divided

P2P technology has three architecture modes: centralized directory structure represented by Napster, pure P2P network structure represented by Gnutella, and hybrid P2P network structure. From the perspective of P2P technology, P2P technology has been divided into four generations: First Generation P2P (Central Control Network Architecture) and second generation P2P (Distributed Network Architecture ), third-generation P2P (hybrid network architecture) and fourth-generation P2P (currently developing P2P technology ).

　　1, The first generation of P2P (Central Control Network Architecture-centralized directory structure)

Centralized directory structure uses a central server to manage P2P nodes. P2P nodes register information about themselves (names, addresses, resources, and metadata) with the central Directory Server ), however, all content is stored on each node rather than on the server, the query node selects and locates other peer points based on the information query in the directory server and network traffic and latency, without going through the central Directory Server. The advantage of a centralized directory structure is that it improves the manageability of the network and makes searching and updating shared resources very convenient; the disadvantage is the Network Stability (if the server fails, all the peer nodes under the server will fail ).

　　2Second generation P2P (Distributed Network Architecture-pure P2P Network Architecture)

The pure P2P network structure is also called the broadcast P2P model. It does not have a centralized central directory server, and each user randomly accesses the network, A group of neighboring nodes establish a logical Coverage Network through end-to-end connections. Content Query and content sharing between peer nodes are directly transmitted through broadcast relays of adjacent nodes. At the same time, each node also records the search track to prevent the generation of the search loop. The pure P2P network structure solves the problem of centralized network structure, and provides better scalability and fault tolerance. Since no peer node knows the structure of the entire network, the search algorithms in the network are carried out in flood mode, the proliferation of control information consumes a large amount of bandwidth and quickly causes network congestion or network instability, resulting in poor availability of the entire network. In addition, such systems are more vulnerable to spam information, it is even a virus attack.

　　3Third-generation P2P (hybrid network architecture-hybrid network architecture)

The hybrid network structure combines the advantages of pure P2P decentralization and centralized P2P quick search. Different Node capacities (computing capacity, memory size, connection bandwidth, and network retention time) are classified into common nodes and search nodes. A self-governing cluster is formed between a search node and several neighboring common nodes. The centralized directory-Based P2P mode is used in the cluster, different clusters in the P2P network are connected to the search nodes through the pure P2P mode. You can select the node with the best performance among each search node again, or introduce a new node with the best performance as the index node to save the information of the search nodes that can be used throughout the network, maintains the entire network structure. Because file search on common nodes is performed in the cluster to which the local node belongs, only when the query results are insufficient can a limited pan-flood be performed between search nodes. In this way, it is extremely effective to eliminate the adverse effects of network congestion and search latency caused by the use of flood algorithms in a pure P2P structure. At the same time, because the search nodes in each cluster monitor the behavior of all common nodes, it can ensure that some malicious attack behaviors can be controlled locally on the network, to a certain extent, the load balancing of the entire network is improved.

　4. fourth-generation P2P (P2P technology in development)

It should be said that the fourth generation of P2P did not form a real generation, but made improvements based on the original technology, and proposed and applied some new technical measures. Typical examples include:

(1) select one of the dynamic ports. Currently, P2P applications generally use fixed ports, but some companies have begun to introduce protocols to dynamically select transmission ports. Generally, the number of ports is between 1024 and ~ In the range of 4000. Even P2P streams can be transmitted using the port 80 (25) originally used for HTTP (SMTP) to hide them. This makes it more difficult to identify P2P streams across carrier networks and master their traffic.

(2) Two-way download. Companies such as eD and BT are further developing and introducing two-way stream downloads. This technology allows multiple parallel downloads and uploads of one file and/or multiple parallel downloads of a part of a file. At present, the traditional architecture requires that the object be uploaded only after full download. This will greatly speed up file delivery.

(3) intelligent node auto-overlapping networks. Smart node auto-overlapping network is a new technology in which the system uses P2P technology to schedule existing IP addresses to carry network resources. In the Router network layer, smart nodes are configured with various link peering connections, an elastic overlapping network at the network application layer. Network security, QoS, and management can be improved while maintaining the Internet Distributed Autonomous system structure. Smart nodes can exchange data between routers and ensure data classification (virus identification and spam. By observing the Internet through multiple geometric distribution nodes, sharing information can help you understand the scope and nature of Internet worm infections. Provides high-performance, scalable, and location-independent message routing to determine the location of the nearest local resource. Improve content delivery. Use smart nodes to detect Internet path traces and send back trace data. This solves the current cross-Autonomous Region path selection problems on the Internet. Implement QoS routing, reduce packet loss and latency, and quickly and automatically restore data.