HPE SSA User Guide: Spare Management Mode clarification.

[parnassus]

Hi all,

I've a doubt related to Spare Management mode's description available on the HPE SSA User Guide (Sheet 31 of 143).

Since I'm implementing a Spare drive to an existing RAID 6 Array on a HPE Smart Array P440/4G Controller I want to be sure I understood correctly the differences between the two described Spare Management modes.

The HPE Smart Storage Administrator User Guide reports two possible Spare drive(s) management modes:

Changing the Spare Management mode

The Spare Management feature provides multiple methods for handling spare behavior.

You can choose from the following options:

• Dedicated: When the failed data drive is replaced, it must be rebuilt from the data on the spare drive. In Dedicated mode, one spare can be dedicated to multiple arrays. [**]
• Auto-Replace Drives:The spare for the failed data drive automatically becomes the replacement data drive. When the spare is replaced, the data drive does not need to be rebuilt. In Auto-replace mode, spare drives cannot be shared between arrays

Shouldn't the Auto-Replace Drives mode (which I tend to prefer over the Dedicated mode) be described this way instead:

"Auto-Replace Drives: The spare drive for the failed data drive automatically becomes the replacement new data drive. When the failed data drive spare is then replaced with a new drive, that data drive does not need to be rebuilt [*]. In Auto-replace mode, spare drives cannot be shared between arrays"

Is my line of reasoning correct?

[*] That's because the Spare drive become (changing its role) the new data drive by replacing the failed one so the RAID Volume doesn't require a second rebuild operation. As a consequence - if I understood it correctly - the drive which replaced the failed one, kicks then in as (or with the role of) the new spare drive (or kicks then in just left unassigned?).

[**] As I understood it means that, with this Dedicated mode, two rebuilt operations happen (the first time when the Spare drive kicks in and the second time when the failed data drive is then replaced with a new good one, this one is the target of the second rebuild operation)...this is quite different from the Auto-Replace Drives mode which limit the rebuild to just one process.

[TTr · Accepted solution]

I like to keep things even simpler

Dedicated: "Fixed hot spare". It is always a hot spare. When a data drive fails, the hot spare is used to rebuilt the failed data drive. When the data drive is replaced, *it* is rebuilt from the hot spare and the hot spare is available again.

Auto-Replace Drives: "Floating hot spare". When a data drive fails the hot spare is used to rebuild the failed data drive and the hot spare is NO longer a hot spare, it becomes a data drive. When the failed drive is replaced, it becomes the (new) hot spare.

No need to talk about shared or not shared between arrays. 

[parnassus]

Thanks, really...your explanation means to me that I basically understood the difference between those Spare (drive) Management modes despite the unclear description provided on the HPE SSA User Guide.

I think I'm going to use the "Auto-Replace Drive" mode...

[TTr]

One other factor for choosing the mode is whether you have a small array with one disk enclosure or a larger array with multiple disk enclosures. With one enclosure you normally choose the "auto replace" mode. It doesn't matter which position the spare ends up after it floats around.

With arrays that have multiple disk enclosure you typically choose "dedicated" mode because you don't want the hot spare in enclosure B, become a data drive for a failed disk in enclosure A.

Also in arrays with multiple enclosures, you typically configure one (or more) hot spare(s) per enclosure and you may allow the hot spare to "float" within the enclosure only. For this the array hardware must have that feature in the configuration options. So this is a new "mode" for the hot spare, it is both "dedicated" and "Auto-Replace". :):)

[parnassus]

Understood. Very reasonable approach.

I think the scenario I'm facing adapts quite well with the Spare Management "Auto-Replace Drives" mode since I have just a 8 LFF Hot-Plug disks cage (lower position) on my HPE ProLiant DL180 Gen9, so 6 disks in use, 2 free LFF bays still free (I initially configured a RAID 6 Volume with first 5 disks installed, the 6th disk was added only yesterday to add the "Spare" functionality and actually it is still into the "unassigned" state...but now I know what to do).

Thanks!

Edit (June 2017): this recently published mmr_sf-EN_US000006614 HPE Knowledge Article tells just half of the story...assuming users will rely on default Spare Management "Dedicated" mode without considering the "Auto-Replace Drives" alternative.