Paralel backup vs normal backup

User_JNHXJ

Hello im learning Rman i have question which one is better paralel backup or Normal backup(only use 1 channel)

what are the advantage of paralel backup ? is it because there is a lot of channel the backup process become faster?

and also does rman read all of the separated backup sets created by paralel backup?

in my opinion normal backup is simple because we have to manage 1 file only while paralel backup create mutiple file which make it harder to mantain.

EdStevens

so what you mean is using more channel is better even at situation?

Even at what situation?

What I'm saying is

1) using more channels (up to a point) will yeild faster performance

2) using more channels does not impose any more admin burden on the DBA.

EdStevens

can i have some clarification about this please?

Perhaps this will help: https://stackoverflow.com/questions/40163095/what-is-socket-core-threads-cpu

EdStevens

Parallel backup is faster for the same reason any other parallel operation is faster than the single thread/channel/path version of the same operation. For the exact same reason that a 6-lane highway can move a given traffic load faster than a 2-lane highway.

in my opinion normal backup is simple because we have to manage 1 file only while paralel backup create mutiple file which make it harder to mantain

Even a single-channel backup can create multiple files, depening on MAXSETSIZE and CONTROLFILE AUTOBACKUP. In any event, it's rman that maintains them, no you, so it is no more difficult to maintain the backups from multi-channel, parallel backups than for single-channel backups.

User_JNHXJ

so what you mean is using more channel is better even at situation?

Dear DBA Frank

In terms of results, running RMAN in parallel or not makes no difference: what is backed up, in how many backupsets , the output volume etc. will not change. With a DEGREE OF PARALLELISM of 8 for instance, 8 CPUs will be involved in your backup, instead of one if you do not run RMAN in parallel (DOP=1), so it should run faster, which should be taken into consideration when backing up high volumes.

Also note that to use PARALLELISM, you must have the Enterprise Edition of the Oracle database -- it's not available with the Standard Edition.

EdStevens

so what you mean is using more channel is better even at situation?

Even at what situation?

What I'm saying is

1) using more channels (up to a point) will yeild faster performance

2) using more channels does not impose any more admin burden on the DBA.

User_JNHXJ

Even at what any situation?*

sorry forgot to update that. also i have question about the cpu. how do i get information of number of cpu in hardware.

thanks

EdStevens

also i have question about the cpu. how do i get information of number of cpu in hardware.

That would be an OS question. And you haven't stated your OS.

So, what did you discover when you googled "how many cpus does my computer have?"

User_JNHXJ

my OS is windows

and my hardware have 1 CPU 4 Cores and 8 logical process

from what i know all hardware usually have 1 cpu. but when i googled it there is a hardware that have mutiple cpu.but my question is when we talking about "CPU" are we referencing the core or actual cpu?this makes me confuse because i read from a blog that an optimal parallelism is based on your hardware cpu.

does that means the optimal parallelism is either 1(CPU) or 4(based on core) and from what i know production database usually run 2 or more degree of paralellism so i assume we talk about the cores of the cpu instead of the actual cpu

can i have some clarification about this please? thanks

EdStevens

can i have some clarification about this please?

Perhaps this will help: https://stackoverflow.com/questions/40163095/what-is-socket-core-threads-cpu