Performance and reliability of enterprise-level Core Switches

Enterprise-level Core switches are worth learning. Here we mainly introduce four key elements of enterprise-level Core switches. Currently, most enterprises in China generally access their networks through the backbone networks of China Telecom. They can only complete some basic information exchange functions, it cannot meet enterprises' complex network application requirements, such as handling diversified businesses, rapidly increasing branches for information exchange, and increasing information exchange volume. Therefore, building a core exchange network has become one of the top priorities of large and medium-sized enterprises.

Enterprise-level Core switches are the key equipment of the enterprise's core exchange network. Considering the general situation of Chinese enterprises at this stage, it is recommended that when selecting enterprise-level Core switches, we should examine the performance, scalability, reliability, and security of our products.

Product performance

Core switches shoulder the "Hub" of information exchange, so they must be a full-line speed, non-blocking switching device. In addition, as the number of ports and load increase, its performance should increase, which requires enterprise-level Core switches to have high design parameters. For example, if the backboard bandwidth is 128 GB, the forwarding rate can reach 96 million packets/second, and load balancing is achieved on two control engines in one chassis, enterprise Users can achieve full-line rate switching at full capacity. In terms of performance improvement, excellent core switching devices are no longer simply pursuing high parameters, but adding more designs close to users.

1. Exchange Structure

At present, there are many different ideas and methods in the structure design of network products. One is to adopt a centralized exchange system, such as putting control engine modules with strong performance and functions on the center, and setting a simple interface card on the interface module, all data is transmitted to the Central card through the Backplane for processing. After processing, the backplane is sent back. The advantage of this design concept is that it can help users Reduce network costs from the ports of each module. The disadvantage is that the port efficiency of network data is relatively low, all data exchanges depend on the data processing board capability of the core board. As the number of boards increases, the performance of the entire switch decreases, that is, the number of ports in the chassis is inversely proportional to the performance.

In addition, the latest distributed switching architecture can be adopted. On each interface card, the local machine processes all data exchange, routing, forwarding, and security filtering, multiple boards are used to process local control and forwarding. The advantage is that the performance is guaranteed. As the number of boards and the number of ports increase, the performance of the entire machine increases linearly. The disadvantage is that each board needs to be added with corresponding processing chips, the cost is higher.

2. bandwidth limit

In the past, most routing methods were based on the target address, while more complex applications needed more advanced policy routing. Taking the establishment of campus network as an example, there are education network outlets, telecommunications departments or operators in the campus network. When deciding what traffic to use and what to export, the school needs to apply a policy route to solve the problem, which is more flexible than the route based on the target address.

Scalability

From the actual situation, almost no network construction in large and medium-sized enterprises is still blank, and most of them are upgrading and transforming the Network, this inevitably puts forward scalability requirements for core switching devices. Is it compatible with existing network products of various vendors and brands on the access end? Can I keep up with the current development trend of network devices and constantly adapt to network development to achieve smooth upgrades?

In 0.1 million, the 10-Gigabit Ethernet standard was introduced, and in the near future, or even Gigabit Ethernet will be released. Therefore, whether the core exchange devices that enterprises have invested heavily in can play different roles in networks in different periods smoothly and stably becomes an issue that enterprises must consider when investing in basic costs.

Reliability

In terms of hardware, can all hardware components be completely based on a design without a single fault point and ensure continuous reliability? Can the software Ensure network reliability and fault tolerance? These are users' concerns. For example, the chassis is designed in accordance with the "Five 9" telecom-level standard, and there is no single fault point. The backboard design adopts the passive backboard structure, that is, the backboard itself is passive. No power supply is required; the power supply adopts the N + 1 Redundancy mechanism, which is redundant in the exchange parallel and control engine. The modules support live hot swapping and so on. These features are of great help to improve the reliability of the equipment. It is worth mentioning that the power supply design. Currently, in many telecom projects or operator projects, 48 v dc power supply is required in many places. In the past, when users in China purchased network equipment, A 48 v dc module usually requires a higher cost.

In addition to the fault tolerance mechanism on the hardware, some measures can be taken on the protocols at different layers to help users improve reliability, such as the layer-2 Spanning Tree mechanism, supporting 802.1W and 802.1 S, this enables faster and more efficient convergence, and implements a layer-2 structure of global load balancing. In Port redundancy, 802.3ad is supported to achieve load balancing between multiple physical ports, redundant backup mechanism for each other; supports VRRP-standard routing redundancy protocol on the third layer, and uses the characteristics of some routing protocols to help users establish multiple equal paths to the same destination, implements load balancing and redundant backup of traffic.

Security

When an attack virus or similar hacker behavior occurs, what role does the network switching device play for network administrators in terms of security features? Some enterprise-level Core switches provide status-based firewalls to help users prevent and reject viruses and some malicious attacks, such as fragment attacks. In addition, some switches also support VPN in terms of security, which can prevent address theft on the port. All security filtering mechanisms can be reflected by hardware. At present, users are most concerned about QoS, and VoIP has the most QoS requirements on the network. With the continuous development of Internet and client applications, the network is becoming richer and richer, and core switching devices must be deployed on the same network hardware platform to meet various requirements, while ensuring the quality of service for users. In this regard, the correct design philosophy of enterprise-level Core switches is to provide different guaranteed bandwidth and service capabilities based on different user consumption conditions.

It is very useful for enterprises that use broadband rates as service commitments to set different service projects based on different users. Core equipment should be designed to ensure the configuration of network equipment according to the enterprise investment, so as to effectively protect the enterprise investment.