13-09-2017, 01:52 PM
A cluster is a group of independent computer systems, called nodes, that work together as a unified computing resource. A cluster provides a unique name for clients to use and a single administrative interface and ensures that data is consistent across nodes.
Windows Clustering encompasses two different clustering technologies. These technologies implement the following two types of clusters.
• A network load balancing cluster filters and distributes TCP / IP traffic across a range of nodes, regulating the load of the connection according to the port rules defined by the administrator.
• A failover cluster provides high availability of services, applications, and other resources through an architecture that maintains a consistent cluster image across all nodes and allows nodes to transfer ownership of resources to demand.
The following are the programming interfaces for Windows Clustering technologies:
• The Network Load Balancing Provider enables developers to create remote configuration and management tools as well as custom user interfaces for network load balancing clusters.
• Failover Clustering APIs allow developers to build cluster-aware applications, deploy high availability for new resource types, and create remote configuration and management tools.
SafeKit software supports as many beats as there are networks shared by two servers. The heartbeat mechanism is used to implement Windows and Linux clusters. It is integrated into the SafeKit cluster with real-time file replication and failover.
In normal operation, the two servers exchange their states (PRIM, SECOND, resource states) through the heartbeat channels and synchronize their start and stop procedures for the application. In particular, in the case of an application failover due to a software failure or a manual operation, the stop script that stops the application runs first on the primary server before running the startup script on the secondary server. Therefore, the replicated data on the secondary server is in a safe state corresponding to a clean stop of the application.
If all the beats are lost, it is interpreted as if the other server was inactive and the local server changes to the ALONE state. If it is the second server that goes to the ALONE state, then there is a failover of the application with the restart of the application on the secondary server. Although not mandatory, it is best to have two push channels on two different networks to synchronize the two servers in order to separate the network failure case from server one failure.
Windows Clustering encompasses two different clustering technologies. These technologies implement the following two types of clusters.
• A network load balancing cluster filters and distributes TCP / IP traffic across a range of nodes, regulating the load of the connection according to the port rules defined by the administrator.
• A failover cluster provides high availability of services, applications, and other resources through an architecture that maintains a consistent cluster image across all nodes and allows nodes to transfer ownership of resources to demand.
The following are the programming interfaces for Windows Clustering technologies:
• The Network Load Balancing Provider enables developers to create remote configuration and management tools as well as custom user interfaces for network load balancing clusters.
• Failover Clustering APIs allow developers to build cluster-aware applications, deploy high availability for new resource types, and create remote configuration and management tools.
SafeKit software supports as many beats as there are networks shared by two servers. The heartbeat mechanism is used to implement Windows and Linux clusters. It is integrated into the SafeKit cluster with real-time file replication and failover.
In normal operation, the two servers exchange their states (PRIM, SECOND, resource states) through the heartbeat channels and synchronize their start and stop procedures for the application. In particular, in the case of an application failover due to a software failure or a manual operation, the stop script that stops the application runs first on the primary server before running the startup script on the secondary server. Therefore, the replicated data on the secondary server is in a safe state corresponding to a clean stop of the application.
If all the beats are lost, it is interpreted as if the other server was inactive and the local server changes to the ALONE state. If it is the second server that goes to the ALONE state, then there is a failover of the application with the restart of the application on the secondary server. Although not mandatory, it is best to have two push channels on two different networks to synchronize the two servers in order to separate the network failure case from server one failure.