reg ASM - External redundancy

[satish]

Dear Erman,

We are planning to use external redundancy diskgroup for datafiles(2node RAC).we would like to understand the below scenarios

external redundancy
using raid 1(mirror at storage level)

1)If some disk was failed(permanent failure/temporary failure),then will we loose any data?
2)Do we need to go for restore in this case?

[ErmanArslansOracleBlog]

When you use external redundancy, you rely on the storage level raid.

So if you mirror your data in storage level, then you can continue your work when 1 disk fails

Important note:  I mean "it can tolerate 1 physical disk fail" -- This physical disk that I m talking about is really a "physical disk" -- the disk drive itself)

When you use external redundancy, ASM won't do any mirroring.. The fault tolerancy in storage layer, become the fault tolerancy of ASM.

External redundancy means -> you use storage array protection features.

[Sri]

Thanks for the update erman,In this case, can we add a new disk and request storage team to sync it

[ErmanArslansOracleBlog]

No.
If a disk fails,  storage team will replace it.. The RAID mechanism will automatically sync the new disk.

In case of a new disk (an additional disk to extend the space) ->  Storage team will add the disk. OS team will make OS see the new disk, then you as DBAs will do the ASM operations.. (adding a disk to ASM)

[satish]

Thanks a lot erman...It clarifies much many things to me

[satish]

Dear Erman,

Need a clarification.

lets suppose,we have a disk in a diskgroup with its header corrupted; the diskgroup was configured as “EXTERNAL” redundancy that means that with one disk inaccessible the whole diskgroup will be dismounted?

Thank you

[ErmanArslansOracleBlog]

ASM related header corruption is another subject.
It is related with ASM.
RAID protects you from physical things.. Not this header corruption.. (actually, the cause of it may be related with a physical failure, but if you are ended up with a corrupted ASM disk header, you should fix it on ASM layer)
ASM Header corruption is a logical thing.. Header is something ASM cares. So you need to repair it from ASM layer. you use kfed and etc to fix it.

Here a document for you ->  How To Restore/Repair/Fix An Overwritten (KFBTYP_INVALID) ASM Disk Header (First 4K) 10.2.0.5, 11.1.0.7, 11.2 And Onwards (Doc ID 1088867.1)

[satish]

very useful erman.Thank you

[ErmanArslansOracleBlog]

In the event of a disk failure with External Redundancy, the ASM Diskgroup becomes unavailable until the disk is made available.

[satish]

Dear Erman,

i am sorry.In the begining of this thread..your said as below

When you use external redundancy, you rely on the storage level raid.

So if you mirror your data in storage level, then you can continue your work when 1 disk fails

Important note:  I mean "it can tolerate 1 physical disk fail" -- This physical disk that I m talking about is really a "physical disk" -- the disk drive itself)

And in your last update it is different "In the event of a disk failure with External Redundancy, the ASM Diskgroup becomes unavailable until the disk is made available"

I request you to let me know which one of the above statement is correct?

[ErmanArslansOracleBlog]

Raid is in storage layer.
OS sees the disks of the storage as luns.
ASM sees these disks/luns coming from OS as ASM disks.
So if your lun is corrupt and if you are using external redundancy, the result will be the same..
Remember, having only the raid (without ASM redundancy) means external redundancy. (from Oracle's perspective)