Re: P9 SSD internal disk options/questions - parity questions

OK I'm being smacked by every editor I've ever written for. I said 'never do no hot spare'. Double negative.

I am ALWAYS in favor of hot spare. Always.

Be warned of the maintenance cost of those 775 SSDs. That will be 'significant'!!

I have moved MANY Partitions by moving the drawer(s). LOVE that approach for simplicity and speed.


- Larry "DrFranken" Bolhuis

www.Frankeni.com
www.iDevCloud.com - Personal Development IBM i timeshare service.
www.iInTheCloud.com - Commercial IBM i Cloud Hosting.

On 3/30/2018 3:33 PM, Steinmetz, Paul wrote:

Larry,

Why no hot spare?

I now remember the rule of two raid sets per controller pair.

It's been too long since the P5 to P7 migration, not remembering all the details.

I'm thinking of using my current production enterprise 775 SSD on my P9 R&D lpar, but the logistics is ugly.

Another thought.
Keep the current enterprise 775 SSD intact in the EXP24S drawer, move the entire drawer to the P9.
This would eliminate a production migration.
Any thoughts, pros/cons.

Paul



-----Original Message-----
From: MIDRANGE-L [mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf Of DrFranken
Sent: Friday, March 30, 2018 3:06 PM
To: Midrange Systems Technical Discussion
Subject: Re: P9 SSD internal disk options/questions - parity questions



I have not heard on the POWER9 CECs what the cache size is but it has been comparable to the EJ14 in the past. The internal RAID cards can control one external drawer in addition to the 18 internal drive slots.

Arms ARE Still important because IBM i expects arms and thus has lower queue depths per arm than some other systems. That said I wouldn't sweat the difference between 20 arms and 40 arms unless you really beat them up hard. The minimum is 6 for a production system with 10 to 18 being a solid base for a fairly busy system.

I would never do 'no hot spare' any longer. Once you hot spare you never go back. One hot spare can be used to back multiple RAID sets on the same pair of RAID cards.

You do want even numbers of RAID sets per controller pair. This allows each card to control one RAID set. Only in a failure then will all drives be controlled by one card. Since you also lose cache in that case performance will suffer, perhaps dramatically.

The minimum RAID set size is 3 for SAS drives but that's very small and wastes close to the same amount of storage as MIRROR yet still incurs the READ READ penalty before writing that mirroring avoids. (Writes to mirrored disk are WRITE WRITE, one to each drive, while RAID5 is READ READ WRITE WRITE one each READ and WRITE to the drive with the data and the parity drive for that data.)


- Larry "DrFranken" Bolhuis

www.Frankeni.com
www.iDevCloud.com - Personal Development IBM i timeshare service.
www.iInTheCloud.com - Commercial IBM i Cloud Hosting.

On 3/30/2018 2:43 PM, Steinmetz, Paul wrote:

I'm doing various P9 SSD configuration options.

One Config is using the CEC with the integrated SAS RAID controller.
What is the amount of cache on the integrated SAS RAID controller?
I'm not finding this anywhere.

Second Config is using an EXP24XS with dual EJ14 - PCIe3 12 GB Cache
Raid Plus SAS Adapter Quad-port

1) Are number of arms still important?.
One config using a smaller SSD, more arms, no room for growth.
Second config using a larger SSD, less arms, room for growth.

2) Did Raid-5 parity set rules remain the same on P9?
One config using 1-Raid5 parity set, no hot spare.
Second config using 2-Raid5 parity sets, no hot spare.

>From help text
All the disk units in a parity set must be the same capacity with a
minimum number of either 3 or 4 units, and a maximum number up to 18
units (depending upon the type of adapter) in the resulting parity
set.

Also, from a performance standpoint, are 2 or 3 raid sets better than 1?
Select parity optimization
1. Availability
2. Balance
3. Capacity
4. Performance

Thank You
_____
Paul Steinmetz
IBM i Systems Administrator

Pencor Services, Inc.
462 Delaware Ave
Palmerton Pa 18071

610-826-9117 work
610-826-9188 fax
610-349-0913 cell
610-377-6012 home

psteinmetz@xxxxxxxxxx
http://www.pencor.com/

--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list To post a message email: MIDRANGE-L@xxxxxxxxxxxx To subscribe, unsubscribe, or change list options,
visit: https://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxx Before posting, please take a moment to review the archives at https://archive.midrange.com/midrange-l.

Please contact support@xxxxxxxxxxxx for any subscription related questions.

Help support midrange.com by shopping at amazon.com with our affiliate link: http://amzn.to/2dEadiD