Hello folks,

I am new for clustering and i am reading documents to understand how it work. But i am always confuse between below technologies.

I would like to know (Actually clear my doubts) about

1) what is fail-over?
2) what is switchover?
3) what is cluster?
4) what is HA?
5) what is Load balancing?

kindly provide short description and when/where i use these technologies.

Thanks

The words you're asking for definitions on can be a LOT of different things, and it's generally down to the conventions of the vendor in question as to what they really mean. I can make some typical generalizations though:

1 - a secondary device taking over the role of a primary one if it thinks that device has failed in some critical way
2 - no idea really... could mean a manual failover, not a specifically common phrase to me
3 - typically a number of devices all actively working as a single device. Again though what constitutes a cluster varies wildly.
4 - all of this. High Availability is a very broad umbrella term that covers all many aspects of service resiliency
5 - spreading a load between multiple nodes. Typically this would be a device sitting in front of a number of back end devices to which application requests or such like, are directed based on various algorithms and policies.

Isn't switchover something like switching applications between hosts? I guess, it is done more often with vps'.

Thanks acid_kewpie, Blue_Ice for your replay

As i understand ( correct me if i am wrong )

HA cluster we can also say fail-over cluster, in this type of cluster we can improve availability of service.i mean if one goes down other take place.

In load balancing cluster we can share load between multiple nodes and improve performance.

one more question can we create HA cluster with load balancing ?

@Blue_Ice
As Wikipedia
Switchover is the manually switch over from one system/server/network to another.Switchover happens with human intervention.

Ok, I wasn't sure about switchover. I only heared about it when talking about clustering virtualization solutions...

Anyway, basically a load balanced cluster (HPC) is also a HA cluster. Because when a node fails in a HPC, the load of that node will be devided over the remaining servers. But HPC's are more complex as you need in some cases active/active clustering with master/master replication. Besides that you would use an extra server that will be the director in your HPC. In a production environment you probably would like the director to be clustered for HA as well.
Take good care of your configuration in a HPC, in the future that will come in handy regarding scalability.

Hmm, I'd suggest that if you can go and find these answers on wikipedia yourself, you maybe should ask in the first place??

I would not directly equate failover and HA clusters. the term clustering *usually* suggests 1) arbitrary numbers of devices (maybe hundreds), not pairs as you'd find in a typical failover and also that there is not as often a backup element, but devices are all actively processing data, and failure of a single device would usually just reduce the processing / throughput capacity until it is restored.

Thanks for clarification. Information provided by both of you, help me in my reading.

I had search out for switchover after my first post.

Thanks,