Load Balancing Technology of Skynet Firewall

1. Load Sharing-solutions for large service nodes

A large amount of access traffic and users are the goal of information service providers. However, a large amount of access will impose a heavy burden on servers. As the number of users of excellent Internet application services increases, the server becomes overloaded. If a large number of user service requests cannot be processed in a timely manner, service interruption may occur. In the past, when solving these problems, we could only use servers with higher computing power to replace the original servers. The old servers could only be eliminated. Even so, the load capacity of a single server is limited and cannot be expanded infinitely. At the same time, the price of high-end servers increases exponentially with the performance of servers, the system model that uses multiple low-cost servers to share loads is becoming increasingly mainstream.

2. Principles of the Load Balancing System

The load balancing system distributes user service requests concentrated on one server to multiple servers. In the early stages of load balancing, many networks were designed to rotate domain names, that is, one domain name corresponds to multiple servers. As a cheap solution, domain Name rotation can solve some server load problems. However, this load balancing method has many limitations: it cannot be based on the load of each server, send user service requests to different servers. When one server fails to work, the system still sends user access requests to the faulty server, some services are interrupted. Because domain name interpretation generally has Cache on servers in various regions, user access requests in one region are concentrated on the same server. As a result, domain name rotation is used to share system load, and the effect is not obvious.

3. Load Balancing System Module of Skynet Firewall

Using the Distributed Solution of Skynet firewall, you can build a large-capacity server cluster system with fast response time and high fault tolerance. The load distribution module of Skynet firewall intelligently distributes user service requests to multiple servers. It also provides Fault Tolerance function to automatically isolate problematic servers. The specific functions of the system are as follows:
　　
1) Dynamic Load Balancing
The Load distribution module of Skynet firewall can automatically select the least Load Server Based on the server Load, including CPU usage and system Load, and send user service requests to the server.

2) fault tolerance Processing
The load distribution module of Skynet firewall can automatically detect server availability. When a server fails, the distributed system automatically bypasses the faulty machine, service requests are not sent to the server, ensuring the normal operation of the system.

4. Working Principle of the Load Balancing module of Skynet Firewall

The load sharing module of Skynet firewall mainly distributes user access to multiple servers according to certain algorithms.
Skynet firewall supports load balancing in different ways. There are two types of functions:

1) Intelligence

Direct Detection
Method: The Server Load balancer module of the Skynet firewall directly sends service request data to the server. Based on the server response time, the server with no response is marked as the problematic server, and the priority of user service request Forwarding is established.
Advantage: it has nothing to do with the system used by the server. The user server can use any type of server.
Disadvantage: The data obtained is inaccurate and cannot be fully shared.

Server Agent
Method: Install the Load detection agent software of Skynet firewall on the server side, and send the server Load information to the Load sharing module of Skynet firewall in real time, including CPU usage, system Load, and network traffic, based on the overall server load index, Skynet firewall establishes the priority of user service request forwarding.
Advantage: It can accurately forward user service requests to idle servers to ensure service quality.
Disadvantage: Server Load balancer agent software must be installed for servers of different operating systems. Currently, only Server Load balancer agent software for Unix systems is installed.

2) fixed class

In a fixed order or immediately forward user service requests to the server. In some special cases, for example, the server does not respond to incorrect service requests or the returned data is abnormal.

5. IIDR Algorithm

In practical applications, Because CGI programs often exist on the server side, these programs store user information in the server's memory. If the server load sharing system cannot identify user sources, the requests of the same user are distributed to different servers, and the program cannot run normally. The load burden module of Skynet Firewall uses the unique IIDR (Intelligent identity recognition) algorithm to ensure that the CGI requests of the same user can be kept on the same server to ensure normal service operation.

6. Skynet firewall Load Balancing Module

Using a distributed structure to build large-scale Internet applications can accommodate a large number of users. However, when the number of users increases to a certain level, the Server Load balancer is in the center of the entire network, it may become a bottleneck of the service system. The SLB module of Skynet Firewall uses a dedicated hashing algorithm to ensure that the system processes a large number of users (more than 30000 simultaneous connections per second, the network efficiency is still above 80%.