Ethernet Technology Encyclopedia (III)

Load Balancing Technology

The proliferation of TCP/IP traffic, new network traffic management equipment also accompanied by the production, such devices provide intelligent content exchange capabilities, the use of this ability to monitor network requests and server systems, through the distribution of access traffic to obtain the best response.

Web, content, or tier fourth to seven-tier switches are no longer unfamiliar, using load-balancing technology to intelligently forward Internet traffic to an application server. More advanced WEB switches can provide traffic distribution based on layer seventh, examine IP packets in more detail, and forward based on HTTP headers, URLs, and Cookies. For global companies, WEB switches can distribute access traffic to servers located around the world, providing users with the best response time and unmatched overall reliability. (Computer science)

SLB Technology

SLB is a generic term that provides the ability to maximize server utilization and provide high overall availability for applications; Transparently realize the load balance of network server, make it to the user, just like a whole; provide manageability. Server load balancing devices use Predictor technology and advanced configurable application combinations to achieve the expected flow distribution results. The Predictor technology is the lower layer technology, which uses the statistical allocation method to divide the data flow of the arrival server.

Further enhancements enable SLB to forward traffic based on higher-level HTTP protocol information. Server load balancing devices are now also known as WEB switches or layer fourth to seven-tier switches.

GSLB Technology

Global server load Balancing or GSLB is a more powerful SLB implementation. It's just that SLB operates in the datacenter, and GSLB is based on the global. Using GSLB not only shortens web response time, but it also makes it possible for customers worldwide to not be aware of server failures.

The basic premise of GSLB is to improve the processing process adopted in the Internet, and to match client requests to the appropriate servers. This can be done by a process called DNS lookup. GSLB The DNS lookup process by arranging the order of IP addresses stored in DNS. Sorting is done based on several metrics that test the health of specific sites that correspond to an IP address. Network processor

Network Processor (network Processor) is a programmable device, which is applied to various tasks in the field of communication, such as packet processing, protocol analysis, routing lookup, Voice/data aggregation, firewall, QoS, etc.

Network processor devices are usually composed of several microcode processors and several hardware coprocessor, and multiple microcode processors are processed in parallel within the network processor, and the processing flow is controlled by a pre-configured microcode. But for some complex standard operations (such as memory operation, routing table lookup algorithm, QoS Congestion control algorithm, traffic scheduling algorithm, etc.), the hardware coprocessor is used to further improve the processing performance. Thus, it realizes the organic combination of business agility and high performance.

NP has the following advantages.

1. Performance

In a hardware platform based on a network processor, a variety of algorithms can be implemented through hardware, the internal general integration of several or even dozens of forward micro-engine and hardware coprocessor, hardware accelerator, in the implementation of complex congestion management, queue scheduling, flow classification and QoS functions, the same can achieve a very high search, forwarding performance, Implement "Hard forwarding."

2. Hardware features that can be flexibly extended

Since NP can support programming, once a new technology or demand appears, it can be easily implemented through microcode programming, the system's "hardware" function can be easily added and deleted by software module (microcode). Therefore, for special user needs, products based on NP can be customized development, that is, can be reduced through the module to meet the needs of different users of products.

So NP provides faster technology, functional follow-up, and more flexible scalability, especially with the support of new specifications and new standards, and products based on NP technology have clearly demonstrated their advantages in the current industry's support for emerging standards such as Mpls/ipv6 and so on.

3. High reliability

NP-based device solutions provide a higher degree of integration, most of which can be implemented using one or two chips, thereby avoiding the pitfalls and functionality and performance degradation of previous implementations through multiple chips and chipset systems, The NP chip system has undergone rigorous testing and various anti-interference and destructive tests before its conversion, thus greatly improving the reliability of NP-based systems.

4. Rich flow classification, congestion management, queue scheduling and QoS features

Most NP uses the parallel operation of hardware, many of the previously used software implementation can not guarantee the performance of the complex policy of QoS, flow operations, and so on, after the use of NP is easier to achieve, at the same time, no impact on performance, in software implementation and based on the ASIC system is difficult to achieve.

5. More convenient and effective management

NP provides the interface with the upper CPU standard or the built-in management CPU, which enables high-speed communication with other CPUs. NP generally provides a large number of hardware counters, can easily achieve a variety of MIB statistical functions, to provide support for network management, but for the business system, there is no overhead, not because of complex, meticulous network management functions affect the performance of the business system.

6. Flexible combinations can be achieved

NP as a device, all provide a flexible configuration function, through the combination of NP and other combinations of other CPUs, to achieve flexible configuration of the system to meet the needs of different equipment, facilitate the system design, speed up the development of equipment progress.

From the current situation, the network processor to achieve the second layer to the seventh layer of programmability and high-end wire speed performance easier said, difficult to do. In addition, writing software code for a Web switch based on a network processor unit is no faster than a design ASIC. The cost savings are less significant because, in many cases, the network Processor unit requires a coprocessor to handle some workloads.