Classification [1] cluster technology
[1.1] High Availability Cluster (Hight Availability, HA)
High availability can be used to maximize the meaning. The main program is to protect business continuity and scalability.
[1.2] cluster load balancing (LoadBalance)
It divided into two parts and the front end load balancer backend service.
Workload scheduling part responsible for the allocation of the client's request to the back-end service nodes, and provide back-end node is really part of the application services.
Different clusters with high availability, load balancing cluster all the back-end nodes are active, they all provide the same service outside, sharing system workloads.
[1.3] scientific computing cluster (High Performance Computing, HPC)
High-performance computing clusters, these clusters is committed to providing powerful computing power of a single computer can not provide, including the numerical calculation and data processing, and a tendency to pursue overall performance. Normally not see this column.
[2] commonly used load balancing Overview
[2.1] commonly used software and devices
Open source software: nginx, lvs, haproxy
Commercial hardware load equipment: F5, Array, Cisco / Huawei and other equipment load
[2.2] commonly used MySQL database load balancing scheme from
(1) F5 hardware load device such - preferably
(2) Keepalived + HAProxy - less preferred
(3) Lvs + Keepalived (LVS connection problems with long defect) - times and selected from
[3] introduced specific technical
【3.1】LVS
(1) concept
LVS: full name Linux Virtual Server, it is a kernel module (ip_vs.so) on LVS nature.
Because LVS load scheduling is implemented in the Linux kernel, the Linux kernel clusters to achieve a high-performance, high-availability load balancing server.
(2) Advantages
> Anti-load capacity, work