The relationship between distributed, cluster, and load balancing

Clustering is a general term. It is divided into several types, such as high-performance scientific computing clusters, load-balancing clusters, and high-availability clusters.

1. Linux clusters are mainly divided into three categories (high-availability clusters, load-balanced clusters, and scientific computing clusters) (only load balancing clusters are described below) Load Balance Clusters Load Balancing Systems: Clusters All nodes are active and they share the workload of the system. General web server clusters, database clusters, and application server clusters all fall into this category.

The load balancing cluster is generally used for the web server and database server of the corresponding network request. Such a cluster can check for fewer, less busy servers and send requests to those servers upon request. From the point of checking the status of other servers, load balancing and fault-tolerant clusters are very close, with the difference being more in number.

2, load balancing system: Load balancing provides a more practical system for enterprise needs. As its name suggests, this system allows the load to be distributed as evenly as possible across multiple computers. The load can be an application processing load or network traffic load that needs to be balanced. In the system, each node can handle a part of the load, and the load can be dynamically distributed between the nodes to achieve balance. The same is true for network traffic. Load balancing has DNS load balancing (commonly used), IP load balancing, reverse proxy load balancing, etc., that is, there are servers A, B, and C in the cluster. They are mutually unaffected and irrelevant. The machine of the station is paralyzed, and it will not affect the operation of other machines. When the user comes to a request, the algorithm of the load balancer determines which machine to process. If your algorithm uses the round algorithm, there are users a, b, c, then processed by servers A, B, and C, respectively.

3, distributed refers to the distribution of different businesses in different places. Cluster refers to bringing together several servers to achieve the same business. Every node in the distribution can be clustered. Clusters are not necessarily distributed. For example: For example, Sina.com has more people to visit. He can make a cluster, put a response server in front, and the next few servers complete the same service. If there is business access, the response server looks at which server is not very loaded. Heavy, which one will be given to complete. Distributed, narrowly understood, is similar to clusters, but its organization is relatively loose, unlike clusters, there is an organization, one server is paralyzed, and other servers can come up. Each node that is distributed has completed different services, and one node is paralyzed, and this service is inaccessible.

Main application of load balancing technology

1. DNS load balancing The earliest load balancing technology is realized by DNS. In DNS, the same name is configured for multiple addresses, so the name is queried. The client will get one of the addresses, allowing different clients to access different servers for load balancing purposes. DNS load balancing is a simple and effective method, but it does not distinguish between server differences and does not reflect the current state of the server.

2, proxy server load balancing Use a proxy server, you can forward the request to the internal server, using this acceleration mode can obviously improve the access speed of static web pages. However, it is also possible to consider a technique that uses a proxy server to evenly forward requests to multiple servers for load balancing purposes.

3, address translation gateway load balancing support load balancing address translation gateway, you can map an external IP address to multiple internal IP addresses, dynamically use one of the internal addresses for each TCP connection request, reach the load The purpose of equilibrium.

4, the internal support of the protocol load balancing In addition to these three load balancing methods, some protocols internally support load balancing related functions, such as the redirection capability in the HTTP protocol, HTTP runs on the TCP connection The highest level.

5, NAT Load Balance Network (Network Address Translation) simply translates an IP address into another IP address, generally used for unregistered internal addresses and legal, obtained Convert between registered Internet IP addresses. It is suitable for solving the situation where the Internet IP address is tight and you don't want to know the internal network structure outside the network.

6, reverse proxy load balancing common proxy mode is the proxy internal network user access to the server connection request on the internet, the client must specify the proxy server, and will send the connection request to be sent directly to the server on the internet Handle to the proxy server. Reverse Proxy means that the proxy server accepts the connection request on the internet, then forwards the request to the server on the internal network, and returns the result obtained from the server to the client requesting the connection on the internet. At this point, the proxy server acts as a server externally. Reverse proxy load balancing technology is to dynamically forward the connection request from the internet to multiple servers on the internal network in a reverse proxy manner to achieve load balancing.

7. Hybrid load balancing In some large networks, we can consider the most suitable load for each server group due to differences in hardware devices, their respective sizes, and services provided in multiple server groups. The equalization method is then load-balanced or clustered again among the multiple server groups to provide services to the outside world as a whole (that is, treating the multiple server groups as a new server group) to achieve optimal performance. We call this approach hybrid load balancing. This method is sometimes used when the performance of a single equalization device does not satisfy a large number of connection requests.