Clustered Hosting
From Website Hosting Wiki
Clustered hosting handles many of the problems that come along with shared web hosting setups. It clusters the handling of security, load balancing and necessary website resources. Clustered hosting is data driven requiring no human involvement when a new account is added to the platform.
Clustered hosting works by virtualizing the physical server’s resources above and beyond normal set limits. By doing this it allows the websites stored on the server to literally access other servers that share processing power so that the application are distributed in real time. Computing power is virtually inexhaustible as the combined total of computer power clustered hosting networks offer websites is enormous compared to the small amount even the largest website will use. Changes to customer accounts are made on every server in the cluster and become live almost immediately where as a shared web hosting server would require a reboot for a change to become live.
This type of web hosting service has tiers of security integrated into the platform. Shared web hosting requires the use of stock solutions that do not always solve issues within the server core and usually only works by integrating the security between the application and the operating system. Firewalls are usually put into place but any weaknesses in the operating system will be exploitable if security problems get through the firewall.
Layer protection in clustered hosting uses a technique called intelligent routing and has redundant switching and a built in firewall and proxy technology. Its security system is more powerful that traditional hosting systems in that it uses mitigating denial of service attacks and network attack solutions, dispersing the attacks over all of the servers. Hardware components that are impacted by these attacks automatically fall out of traffic handling during the attack, but this does not slow down the server or access.
Clusters can also come in what is known as high-availability clusters. These clusters are put together for providing services at a high level and have redundant computers or nodes that back up each other when one fails. For example, if an application crashes on one server, another server within the clusters node will ‘step up’ and take over. No administrative intervention is required. The process is automatic and is known as ‘failover’. The software may configure the cluster node by importing file systems and other things that it will need to run properly.
High-availability clusters are primarily used for running critical databases, file sharing, business application, and customer services related to e-commerce websites. They have redundancies built into them that help eliminate single failure points. Because they use what is called a heartbeat private network connection to monitor the health and status of each node in the cluster, it needs to be able to handle split brain functions. Split brain happened when all of the private links in a cluster go down, but the node is still working. The cluster must be able to handle this and recognize that only certain things have gone down or else it will attempt to start services that are already running and cause data to be corrupted.
