Hybrid Web Cluster: Difference between revisions
m repaired link(s) to disambiguation page - (you can help) |
m WP:CHECKWIKI error fix for #61. Punctuation goes before References. Do general fixes if a problem exists. - using AWB (8853) |
||
| Line 17: | Line 17: | ||
}} | }} | ||
'''Hybrid Web Cluster''' is a software product developed by Hybrid Logic Ltd., a [[London]], [[United Kingdom]] based company. The software is designed to provide commodity web hosting in a distributed and fault-tolerant manner <ref>[http://www.youtube.com/watch?v=kauUyb63Sak Hybrid Web Cluster video demo]</ref> on a [[ | '''Hybrid Web Cluster''' is a software product developed by Hybrid Logic Ltd., a [[London]], [[United Kingdom]] based company. The software is designed to provide commodity web hosting in a distributed and fault-tolerant manner <ref>[http://www.youtube.com/watch?v=kauUyb63Sak Hybrid Web Cluster video demo]</ref> on a [[Computer cluster|cluster]] of either real physical servers or virtual [[Cloud_computing#Infrastructure|cloud computing infrastructure]] server instances. | ||
==Features== | ==Features== | ||
* Runs standard [[ | * Runs standard [[LAMP (software bundle)|LAMP stack]] web applications. | ||
* Works on real hardware and [[Hardware virtualization|virtualized]] [[Cloud_computing#Infrastructure|cloud computing infrastructure]]. | * Works on real hardware and [[Hardware virtualization|virtualized]] [[Cloud_computing#Infrastructure|cloud computing infrastructure]]. | ||
* Has a configurable level of replication redundancy. | * Has a configurable level of replication redundancy. | ||
| Line 26: | Line 26: | ||
==Technology== | ==Technology== | ||
This software splits each website and database into individual units of work - the load generated by each is monitored closely and the cluster automatically and transparently transfers websites and databases between cluster nodes such that no single node is ever overloaded. A single website can go from having no load at all to requiring its own dedicated web and database servers within seconds. Only one server is ever live for a particular website or database at any one time, enabling existing applications to run un-modified just as they would on commodity [[ | This software splits each website and database into individual units of work - the load generated by each is monitored closely and the cluster automatically and transparently transfers websites and databases between cluster nodes such that no single node is ever overloaded. A single website can go from having no load at all to requiring its own dedicated web and database servers within seconds. Only one server is ever live for a particular website or database at any one time, enabling existing applications to run un-modified just as they would on commodity [[Shared web hosting service|shared web hosting]]. | ||
==Scalability== | ==Scalability== | ||
The upper bound on the [[scalability]] of this system is that no single website can ever utilise more than two servers at once - under this condition all standard LAMP applications will run unmodified. However, it is possible to scale beyond this limit via the use of [[Multi- | The upper bound on the [[scalability]] of this system is that no single website can ever utilise more than two servers at once - under this condition all standard LAMP applications will run unmodified. However, it is possible to scale beyond this limit via the use of [[Multi-master replication|multi-master database replication]] - this requires that applications store all [[State variable|variable state information]] within a database - the database can then be live on many servers, the web application itself must remain static and can be served from as many nodes in the cluster as required. | ||
==Redundancy== | ==Redundancy== | ||
It is possible to specify a '''redundancy invariant''' - this sets the number of machines in the cluster that should hold a copy of each website or database. By default this number is set to 2, this means that under normal operation all data will be held on at least two cluster nodes at any one time. If a cluster node fails, the system acts to automatically recover its data and resume web hosting within 15 seconds. This is achieved using the filesystem snapshotting capabilities of [[ZFS]] <ref>[http://www.ukoug.org/calendar/show_event.jsp?id=5122 London OpenSolaris user group talk]</ref> | It is possible to specify a '''redundancy invariant''' - this sets the number of machines in the cluster that should hold a copy of each website or database. By default this number is set to 2, this means that under normal operation all data will be held on at least two cluster nodes at any one time. If a cluster node fails, the system acts to automatically recover its data and resume web hosting within 15 seconds. This is achieved using the filesystem snapshotting capabilities of [[ZFS]].<ref>[http://www.ukoug.org/calendar/show_event.jsp?id=5122 London OpenSolaris user group talk]</ref> | ||
==Release== | ==Release== | ||
Hybrid Web Cluster is scheduled to begin beta testing in September 2010<ref>[http://www.hybrid-cluster.com/#/blog/2010/08/press-release-beta-testing-due-to-begin-in-september/ Press release: Beta testing due to begin in September]</ref> | Hybrid Web Cluster is scheduled to begin beta testing in September 2010.<ref>[http://www.hybrid-cluster.com/#/blog/2010/08/press-release-beta-testing-due-to-begin-in-september/ Press release: Beta testing due to begin in September]</ref> | ||
== See also == | == See also == | ||