Amount Of disk used for redundancy

[sidi_2]

Hi.
I donâ t understand
How the amount of disks used for redundancy is calculated ?
Is it possible to reduce this quantity ?


You can view the result of armdsp â a commande in the attached file.

Th

[Sanjay_6]

Hi,

Maybe this will help you with the calculation. The available capacity depends on the raid level used.

http://www.pcguide.com/ref/hdd/perf/raid/levels/comp.htm

Hope this helps.

Regds

[Mike Naime]

The amount of overhead lost to redundancy is dependent on the RAID technology, and your RAID level. It also depends on how you want to present the data.

Example:
In an HSG80 storage array, With an 18GB mirrorset that uses 2 physical spindles. I am using 36GB of RAW storage. I have 18GB of Usable space, and I loose 18GB to redundancy.

Now here is the question: Is this 50% or 100% lost to redundancy? It is 50% of the total raw space. (One perspective) But it is equal to 100% of the Usable space. (A second perspective)

A related question is "What % is your Usable space?"


Mike

[James C. Nix]

Define redundant.

In an HSG80 Raid 0+1 configuration five 18GB disks will provide a 90 GB striped set for performance. Mirroring, to provide redundancy, will require 5 additional 18 GB disks. For maximum availability at least one 18 GB disk is needed in the spare set for automatic failover. In the event of a failover, an additional disk is needed in stock to hot-swap out a failed disk while retaining the spare.

That totals ten disks , with 216 GB of storage providing 90 GB of high availability capacity. Other RAID configurations would require fewer disks but have different trade-offs that must be addressed in a DR plan.

Throw in a redundant HSG80, independent BA356 racks, redundant power supplies per rack, a VMS cluster (minimum 3 nodes), UPS and diesel generator and years of uptime can be achieved with prudent maintenance.

The "raw" capacity of a disk is never achieved in practice. The % lost to overhead is configuration/OS dependent.

[sidi_2]

Actually I use AutoRAID. For the greatest part of the data it must use RAID
5DP and not RAID 1+0 i.e. I must lose one disc among four whereas currently I loses
two, it seems that AutoRAID puts all in RAID 1+0,
It is what I do not understand


Thaks for your responses

[Roger Buckthal_2]

Yes, the VA is far more complex.

(note many people call the Model 12H, "AutoRAID". The 12H was the predecessor to the VA. What you're asking is about the VA in AutoRAID mode.)

Because the VA is so flexible in the number and size of disks it can simultaneously handle, the ability to add any number or size of new disks to the array, and the active hot spare, the usable capacity computation is very complex.

In the AutoRAID mode, the basic logic is; 85% of the physical capacity is RAID 5DP, 10% is RAID 1+0, and ~2-5% is reserved (reserved free space and metadata space). The physical to usable for RAID 1+0 is 50% of the physical. The percentage for RAID 5DP varies by the number of disk in the RAID group, but itâ s an n+2 organization. For example, 10 disks in the RG, 8 data, 2 redundancy, thus 80% usable.

Itâ s the same basic logic, but when you have an odd mixture of 18, 36, 73â ¦ disks the logical get very complexâ ¦

Then add the active hot spare logic; which does not reserve a physical disk, but rather the virtual capacity to complete a rebuild. The AHS reservation logic knows it can convert some of the RAID 1+0 to RAID 5DP to complete the rebuildâ ¦.. itâ

[Ted Buis]

Do you use AutoRAID on the Model 12 or 12H or on the newer VA7xx0 series?