Disk Enclosures
1752725 Members
5523 Online
108789 Solutions
New Discussion юеВ

understanding Auto-RAID disk allocations

 
Jim Wolff
Occasional Advisor

understanding Auto-RAID disk allocations

Hi all, I have been having a hard time understanding the algorythym HP-UX and AutoRAID uses for my own planning purposes of purchasing new disks.

Recently when I add large disks to an already populated disk array of smaller disk, nearly 80% or more of the entire volume of raw disk space I add to the disk array is consumed in the redundancy and hot swap.

I understand that RAID-05 will automatically take out the equivalent of an entire disk space equal to the largest disk for purposes of HOT-SWAP, but is there any way to control how much space is taken out for the RAID-01 redundancy (mirroring)?

For future planning purposes, what algorythym does HP-UX AutoRaid use? Any white papers on this issue?

I already realize that mix-matching sizes of disk is not the best choice and will not give us what we expect to see when adding larger disks, but when we already have our 12H disk arrays full of 9-Gb drives, it is not always economically possible to completely replace them all with the larger 36-Gb drives all at once.

Thanks in advance.
Sr. Oracle DBA
2 REPLIES 2
Sandor Horvath_2
Valued Contributor

Re: understanding Auto-RAID disk allocations

Hi !

I doesn't know much from AutoRaid but
You can't control directly RAID1 - RAID5 space. It depends disks space and usage.
You can check it use the next command :
arraydsp -v $(arraydsp -i | awk ' { print $NF } ')

regards, Saa
If no problem, don't fixed it.
Mark van Silfhout
Trusted Contributor

Re: understanding Auto-RAID disk allocations

Hi Jim,

I have attached a whitepaper for you, here's a quote from it:
With traditional disk arrays, the performance of the subsystem varies with the configured RAID level and the system workload. As an example, sequential operations may be more efficient in one RAID level verses another RAID level, or random reads may be more efficient than random writes within the same RAID level. HP AutoRAID technology
understands these differences and dynamically adapts its algorithms to best meet the demands of the host.
---- EOQ
You wrote you've added the large disks (not replacing smaller ones), so it will take a while before the autoraid actually starts using them. When you've auto include enables the autoraid will automatically include the disks in the configuration. However the 'balancing proces' takes places when there's almost no I/O (typically at night, or during weekends). This process will copy pieces of the original disks to the newly added disks and spread all data across all known disks. So after a few days you should notice that your working-set-size increases and your raid 0/1 redundancy should decrease (as a percentage of the raw disk capacity). This process can take up to 7-8 days (depending on the I/O load).