We have a 1.5 TB of data (nonasm) and we plan to move into ASM by choose normal redundancy.

Normal Redundancy ==> Two Way Mirroring , So we need 3TB data storage for 1.5TB data including mirroring(SAN disk).

Please clarify below:

A) If we choose single RAW disk 3TB.

How many disk group we can split?

either each 500gb into 6 parts? (1 disk group)

This will be efficient for asm operation?

If the failure group we need to specify mandatory?

In our case, we have a single 3TB disk for storage, so no need to specify failure group right?

From blogs,

((For small numbers of disks (<20) it is usually best to use the default Failure Group creation that puts every disk in its own Failure Group))

so It can take by default itself. So no need to specify right?

B) If we have separate two disks each 1.5 TB

How many disk group we can split, either each 500gb into 3 parts? (Total 2 disk group a and b)

If the failure group we need to specify mandatory?

Shall we specify manually failure group of A disk group is B and failure group of B is A?

Its a better performance?

If you cause a failure group specification it causes a spindle failure?

Either A or B case, which one is better performance approach for I/O operation in ASM?

  • I don't understand your requirement. Could you clarify if you either have one physical or SAN disk and plan to partition it into several partitions? What kind of performance impact do you expect?
    – jmk
    Apr 4 '19 at 17:25
  • We choose single SAN disk 3TB for ASM, we need to create a disk group. For that we need to split into partition into every 6 parts 500 GB ? or we can divide into two parts each 1.5TB?
    – ram kumar
    Apr 5 '19 at 4:54
  • Why do you carve up your single SAN Disk? Do you try to create your OCR diskgroup? For any other diskgroup you can use external redundancy if you reley on your SAN doing the redundancy for you. Partition SAN disks and mirror on the same disk is pretty useless.
    – jmk
    Apr 5 '19 at 6:08
  • yes mirroring and san disk is in same disk group means waste , for that case we use external redundancy for performance
    – ram kumar
    Apr 5 '19 at 9:13
  • One more doubt is we chosse a single disk and of 3TB and chosse a normal redundancy by dividing into two parts means it not useful right ?
    – ram kumar
    Apr 5 '19 at 9:14

ASM is not magic.

Each disk has a maximum number of IOPS it can support. Each disk has a maximum throughput. The more spindles (HDDs) you have, the more I/O you can get. If one disk decides to take a permanent vacation and you don't have data on another disk.... I really hope you have viable tested backups.

From my tests, partitioning a single HDD into multiple LUNs and then using those LUNs for ASM will have a negative effect (albeit, ~-1%).

If you want performance, you will buy 6+ 500GB HDD (or SSD). Each disk will have 1 LUN - the whole thing. ASM will take care of the stripe and mirror.

  1. You have a storage with several disks.

  2. You then group them in a single logical entity which is a disk pool, or whatever it is called by the vendor of the storage.

  3. Then you create multiple small logical entities called virtual disks, volumes, whatever, then present them to your host.

  4. Then again, you group these small disks again into a single logical entity called an ASM diskgroup.

What's the point of step 3 + 4? Whats the point of logically splitting your disks into small virtual disks then grouping them again?

Just create 2 x 1.5TB disks, done.

  • There might be a benefit in using multiple smaller disks/LUNs if you have high I/O loads as you have more I/O queues you can work with. But this depends highly on your OS and storage infrastructure
    – jmk
    Apr 4 '19 at 17:28
  • @jmk Multiple smaller disks, yes I agree. Multiple smaller LUNs, I am not convinced. Apr 4 '19 at 18:14
  • I'ts a thing on SANs, too. Granted it's very dependent on your SAN/OS. In my case multiple LUNs where the solution to overcome issues with the SAN virtualization.
    – jmk
    Apr 5 '19 at 10:22

I think you should read the Oracle Admin Guide.

One of the points there:

The number of LUNs (Oracle ASM disks) for each disk group should be at least equal to four times the number of active I/O paths. For example, if a disk group has two active I/O paths, then minimum of eight LUNs should be used. The LUNs should be of equal size and performance for each disk group.

An I/O path is a distinct channel or connection between storage presenting LUNs and the server. An active I/O path is an I/O path in which the I/O load on a LUN is multiplexed through multipathing software.

Please note, that in older versions of the documentation a minimum of four disks/LUNs per disk group was recommended, so 18c is more strict.

You need to decide who is doing the redundancy. If your SAN/Storageteam does it, then use external redundancy.

Partitioning a disk (SAN or local) and feeding these partitions to ASM is really a bad idea for real servers, but it's fine for playing around within the desktop virtualization of your choice.

  • Ok Thank much for clarifying above.
    – ram kumar
    Apr 5 '19 at 10:54

We choose a 3TB for a diskgroup.(Normal redundancy for 1.5TB data)

Below disk range:

1.5 TB * 1 = 1 Disks 500 GB * 3 = 3 Disks

For an even size, we can give as 500 Gb for the diskgroup consist of 4 disks. So in this case, we split one 1.5TB disk into 3 parts and will provide a normal redundancy.

we did not split all the 4 disks. 3 disks are as same, only one disk we split into parts for the same size. Is this ok?

Please confirm.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.