High speed is a relative concept, faster than fast is high speed. Therefore, based on the current fast Ethernet used in local area networks (LANs) is 100M, and a transmission rate greater than 100M can be considered a high-speed local area network. The technologies that have been adopted in this regard are mainly Gigabit Ethernet and ATM, and the ones under trial are 10G Ethernet. Due to its advantages such as low cost of ownership, good interconnectivity, and many manufacturers, Gigabit Ethernet has actually become the mainstream technology of high-speed LANs.
Technology to build high-speed LAN
Wiring technology
Most of the network wiring currently installed are non-shielded twisted pair wires, and the standards they follow are generally the "Super Class 5" standards announced by EIA/TIA and ISO (of course, the earliest wiring was not met). These performance standards can meet the requirements of Gigabit Ethernet and asynchronous transmission modes at speeds above 1.2Gbps. The predetermined Class 6 cabling frequency limit is 200MHz, so it is hard to say how high the rate will be achieved in future coding systems operating at up to 200MHz. Therefore, in the construction of LANs, data rate is not the decisive factor in the shift to optical fibers.
The cost comparison also illustrates the main reason why unshielded twisted pair continues to be the medium selection in the horizontal channels connected to the workstation. It is obvious that the cost of fiber to desktop is much higher than the cost of unshielded twisted pair. Generally speaking, the cost of passive components in the former is more than three times that of the latter. If the cost of active equipment is added, such as hubs and network interface cards (NICs), the cost difference will be further increased.
However, distance limitation is an inevitable choice for floor connections and internal interconnections in the park. In addition, the explosive growth in bandwidth demand requires network cabling to take into account future smooth upgrades. Therefore, in structured wiring, since the installation conditions of the backbone are generally very difficult, network planners must consider using cables with the highest capacity; in the construction of park networks, fiber to the community and fiber to the building are generally required.
On the other hand, the cost of fiber wiring is declining significantly. This makes multi-mode fiber and single-mode fiber have a high cost-effectiveness. Composite cables are now being installed in many buildings, namely, multi-mode fiber and single-mode fiber are used at the same time. This represents a new development trend and is worth referring to.
Link layer technology
Gigabit Ethernet can provide 1Gbps communication bandwidth and is simplified by Ethernet. It adopts the same CSMA/CD protocol, the same frame format and the same frame length, and also supports full duplex and EtherChannel. For the majority of network users, this means that existing investments can be extended to Gigabit Ethernet at reasonable initial overhead. In this way, Gigabit Ethernet can smoothly transition on the basis of current Ethernet, comprehensively balance various factors such as existing endpoint workstations, management tools and training foundations, resulting in very low overall overhead, and is the preferred technology in LAN construction at present.
The physical layer of Gigabit Ethernet, like Ethernet and Fast Ethernet, only defines the physical layer and the media access control layer. In terms of implementation, the physical layer is the key component of Gigabit Ethernet, and three transmission media are defined in IEEE802.3z: multimode fiber, singlemode fiber, and coaxial cable. IEEE802.3ab defines unshielded twisted pair media. In addition to the above transmission media, there is also a standard 1000Base-LH defined by multiple manufacturers. It is also an optical fiber standard with a transmission distance of up to 100 kilometers. Another feature of the gigabit Ethernet physical layer is the use of 8B/10B encoding, which is the same as Fiber Channel technology. The same benefit is that network equipment manufacturers can use existing 8B/10B encoding/decoding chips, which will undoubtedly shorten the product development cycle and reduce costs.
Multi-layer exchange technology
Currently, switching technology can be divided into second layer switching and multi-layer switching, but strictly speaking, switching means the connection between the source and the destination address. No technology above the second layer can be said to be an exchange technology. Load balancing has largely replaced the term Layer 4 exchange, just as the term application cognition has largely replaced the term Layer 7 exchange.
The second layer exchange is the exchange of the second layer of OSI or MAC layer. This is the switch in our usual sense, which is technically very mature. It works on the second layer of the OSI layer 7 model, namely the data link layer, and the switching is based on MAC addresses.
The third layer switching, or network layer switching, is located in the third layer of the OSI protocol, and it provides higher-level services, such as routing functions. In the past, routers usually used software to achieve network interconnection, but routers were expensive and forwarding slow, becoming increasingly a bottleneck in the network. Layer 3 switching is to integrate routing functions into switches with the help of line-speed switching technology. This switch is called a routing switch or a layer 3 switch. The third layer switching can realize line-speed switching at all network levels, and the performance has been greatly improved. At the same time, it retains the network topology and services on the third layer. These structures and services have great advantages in network segmentation, security, manageability, and broadcast suppression. The goal of the Layer 3 switch is to replace existing routers, which provide information flow communication between subnets, increasing communication speeds from hundreds of packets per second to millions of packets per second. Layer 3 switching is designed to forward multiple protocols at high speed, or provide firewalls to protect network resources, or enable bandwidth reservations. The backbone switches of the LAN will be all third-layer switches.
The fourth layer exchange technology uses information in the third layer and fourth layer packet headers to identify application data flow sessions. Using this information, the Layer 4 switch can make intelligent decisions about where to forward the session transport stream. By doing this, user requests can be forwarded to the "best" server according to different rules. Therefore, the fourth layer switching technology is an ideal mechanism for transmitting data and implementing load balancing between multiple servers.
At present, many products have supported multi-layer exchange, such as Cisco Catalyst 5509/6509, Extreme Diamond series, Foundry BigIron series and Alteon ACE-180e.
Multi-layer switching technology has now been described as an integrated and complete solution that can support various LAN architectures, which intelligently and organically combines switching technology and routing technology. Multi-layer switching technology combines the optimal characteristics of LAN switching technology and routing technology, has higher performance-price ratios and stronger flexibility than traditional router-based LAN backbones, and is the basis for high-speed LAN implementation.
Issues to consider
The networking mode of high-speed LAN is now very simple, and there is no better choice: basically Gigabit Ethernet is the backbone, and high-performance layer two and three switches are used as the core; in terms of network wiring, it is recommended to use multi-mode or single-mode fibers between the backbone and switch, and horizontal wiring can use more than five types of unshielded twisted pair wires. As mentioned earlier, this structure is easy to expand and upgrade. Switch products include Cisco 6509/6509 OSR, Foundry BigIron 8000/4000, Extreme Black Diamond 6816/6808, Alcatel PowerRail 5200/2200, Lucent Cajun P880, Riverstone RS32000/RS8600, Huawei MD5500, Julong RS6006G/RS6004G, Chuangxiang AR8000, etc.
However, whether a network is successfully built must also be considered:
Business development
Whether the business can be carried out and whether the functions are limited are the main criteria for various technical evaluations. Now, the built high-speed information networks are required to be targeted at comprehensive services including voice, video and data. Therefore, whether to support various VLANs and IP multicasting is an issue that must be considered when selecting products.
Mature technology
Computer network technologies including Gigabit LAN and high-speed routers actually have problems with incomplete control domains. There will be many insurmountable obstacles when conducting critical data services or continuous media information communication. Whose product is good or bad, there must be examples to prove it, and it cannot be left to the product's own documentation or the manufacturer's boast. Do not use immature network technology easily, this is the basic principle for establishing high-speed LANs for key services.
Network interoperability
Network interoperability is the most important manifestation of network value. Network interoperability is not only reflected in the geographical coverage area, but also in the interconnection with other networks. The interoperability of high-speed LANs is mainly reflected in the interoperability with the original network and the interoperability with higher-level networks.
Network reliability
Network reliability must be supported through network protocols, device backups and routing backups. In particular, it is very important whether the control and management system of the network protocol itself has high reliability.
Article entry: csh Editor in charge: csh