I think it’s time for a lab experiment with hybrid aggregates of different speeds. Maybe there’s a way to mix in some 7.2k drives for fun.

.

On Nov 27, 2017, at 10:19 PM, Jeffrey Mohler <jmohler@oath.commailto:jmohler@oath.com> wrote:

A storage joke, funny. ;) As much as I named my 2nd 928 'Spare Drive'...

But really, mixing is a BEST practice not-to-do, but when you explore the practicability of doing so, it's not so bad, and the differences exist only on paper* when you do whats said here, 2.5" in a 3.5" aggr.

Its a lot of fun math, but sector density per RPM...etc...10k is the "same" as 15k. Smaller disk, shorter seeks (butterfly worst case)....there's really no argument that supports that this is a bad idea anymore.

Because the best-practice back when, was when everything was equal.

It no longer is..so it doesnt matter in the context of the old maths anymore.

On Mon, Nov 27, 2017 at 12:48 PM, Francis Kim <fkim@berkcom.commailto:fkim@berkcom.com> wrote: After years of being told to not mix drive speeds, this all sounds like a lot of spinning.

.

On Nov 27, 2017, at 9:41 PM, Tim McCarthy <tmacmd@gmail.commailto:tmacmd@gmail.com> wrote:

The rotational latency is also somewhat negated as the platters are also smaller.

The decrease in physical size from 3.5 to 2.5 offsets some of the 15k to 10k difference.

There are lots of write-ups about 2.5 10k vs 3.5 10k drives.

Spelling errors courtesy of Swype for iOS...

Tim McCarthy, Consultant at Presidio ________________________________ From: toasters-bounces@teaparty.netmailto:toasters-bounces@teaparty.net <toasters-bounces@teaparty.netmailto:toasters-bounces@teaparty.net> on behalf of Francis Kim <fkim@BERKCOM.commailto:fkim@BERKCOM.com> Sent: Monday, November 27, 2017 3:12:49 PM To: John Adams Cc: toasters@teaparty.netmailto:toasters@teaparty.net Subject: Re: 3.5" 15KRPM -vs- 2.5" 10KRPM SAS

Shorter stroke is true enough but there’s no getting around the longer rotational latency, which determines the bulk of the HDD access time.

.

On Nov 27, 2017, at 9:07 PM, John Adams <intheyc@gmail.commailto:intheyc@gmail.com> wrote:

Hello, I have AGGRs made out of 3.5" 15KRPM SAS drives. One of these drives failed, and NetApp sent a 2.5" 10KRPM SAS drive in a 3.5 sled as replacement.

After asking about this, they said:

After investigating, the following were my findings:

"As part of the NetApp qualification process, there is no performance degradation with the qualified X90-412B-R6 / SP-412B-R6 10K RPM drive as the seek distances are reduced to be equal to a full stroke 15K RPM drive."

As for the physical size difference:

"The only difference is that the 412A is a 3.5 inch HDD while the 412B is a 2.5 inch HDD. The difference in size is negated by a bracket within the carrier assembly in the 412B. So while the two HDD’s look different, the 412B is fully compatible with the 412A."

So what's your take on this? The "don't mix 10K and 15K drives in the same aggr" keeps ringing away in my head.

Thanks for your thoughts.

_______________________________________________ Toasters mailing list Toasters@teaparty.netmailto:Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters

_______________________________________________ Toasters mailing list Toasters@teaparty.netmailto:Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters