HA Cluster

a a a

HA Cluster are used to raise the availability resp. enhance fail safety. In the case of a failing cluster knot the services will be allocated to other knots. The following subchapters are about such clusters.

With Happyware HA Cluster you will get an even higher availability of your servers and services. Together with you we analyse your actual situation. On this base Happyware develops a custom tailored cluster solution, optional with Microsoft or with Linux.

Up to 15 servers can be attached to a HA Cluster. Depedening on the demand also file systems can be syncrhonised via the network. So it becomes possible to store identical data on several servers. Application examples: Data Base Server, Mailserver and Fileserver.

Shared-Nothing System

Generally HA Clusters are distinguished in two kind of architectures, on the one hand the Shared-All System and on the other hand the Shared-Nothing System.

In a Shared-Nothing System each knot works on its own and independently. It is considerably that operations have to be done with an own processor and respective memory (HDD and RAM). Shared-Nothing Systems are popular because of the independance of their data, e.g. in the field of Mailservers and Metadataservers.

Shared-All System

The second architecture of HA Clusters is the Shared-All resp. following the Stonebraker Classification the Shared-Everything Architecture. The difference to a Shared-Nothing System ist hat the hard disc storage unit is used together by all processors.

Active / Passiv

The Active / Passive Cluster is also named as Failovercluster. Hereto are at least two systems required. The objective is that the second computer takes ofer all tasks resp. the whole system (Failover) in case the first computer breaks down. In this case active mans, a system does the work actively. The active system works, the passive system waits.

Naturally the Standby Systems can meanwhile do other work. A Cluster Mnagement Software is needed to manage this in connection with being standby for a failover.

Active / Active

In contrast to the above mentioned cluster all systems are working active. The Active / Active Cluster is a junction of computers. If one system fails, the other active cluster takes ofer all duties resp. the whole system. The advantage of this configuration is that the complete system works permanently. A failure would only have an impact on performance.

Cluster Dateisystem DRBD

DRBD is an Open Source Network Storage Solution. As shown in the above picture the DRBD is said to be a RAID 1, based on a network. The pictured servers form a typical HA Cluster. Here are the typical components of a Linux Kernel included. A block device is mirrored on a productive server and in realtime on another server. This means the whole systems has always an identical data content on both plates.

DRBD® and the DRBD logo are trademarks or registered trademarks of LINBIT in Austria, the United States and other countries.

Ihr Ansprechpartner

Cluster Experte Jürgen Kabelitz bei Happyware
Sie haben Fragen zu Clusterlösungen?

Unser erfahrener Cluster-Experte Jürgen Kabelitz unterstützt und begleitet Sie bei der Planung bis zur Realisierung Ihrer Projekte.

Angebotsanfragen sind selbstverständlich unverbindlich und kostenlos.

Phone: +49 4105 77000 79
E-Mail: JKabelitz@happyware.com