Load Balancing and redundancy are often used for key services and large-capacity services.
I. Load Balancing:
This solution is required for services with large loads and large capacities. It distributes loads evenly to load each service.
There are many network service load balancing solutions:
1. vrrp-based network service: the Key to this solution is to use the ARP Protocol. More than two machines regularly tell the Access Gateway that they are a service IP address, the Service will act as the IP address of the machine according to the load, or regularly, so as to balance the load, but the business data here should be consistent. The related sessions must be synchronized in a unified manner.
2. network hardware-based services: they are located at the entrance of the service network and connected to hardware devices. Some network devices forward business data to the corresponding servers. Business Data Synchronization is required.
3. Service-Based Load Balancing: each server performs service for a specific ID segment by Business ID segmentation/domain division. This requires unified management of servers.
4. database load balancing: This is complicated. MySQL has a master-slave synchronization server solution and a server family solution. Oracle has its own cluster mechanism and load solution. 10 years ago, oracle load is used. DB2 also has its own cluster solution share-nothing, which has been mentioned on the Internet and requires third-party software.
There are also many redundant backup solutions:
1. vrrp-based: one master node and multiple slaves. When one server stops the service, vrrp automatically implements the redundancy function because the machine that stops the service does not send ARP packets to compete for control.
2. Business-based: There is a Server Manager, which indicates that the managed server is valid upon heartbeats. When there is no heartbeat timeout, the management server will not register the managed server, or start the backup server.
3. Storage-based: everyone should be familiar with raid Redundant Array solutions. Not to mention.
4. redundant backup solutions for databases are too diverse and will be specially organized later.
5. Proxy-based. This principle is basically the same as the business-based approach.