public inbox for gentoo-user@lists.gentoo.org
 help / color / mirror / Atom feed
From: Laurence Perkins <lperkins@openeye.net>
To: "gentoo-user@lists.gentoo.org" <gentoo-user@lists.gentoo.org>
Subject: RE: [gentoo-user] Backup program that compresses data but only changes new files.
Date: Mon, 15 Aug 2022 21:11:19 +0000	[thread overview]
Message-ID: <MW2PR07MB405832F3410A6FFA2C67BA5BD2689@MW2PR07MB4058.namprd07.prod.outlook.com> (raw)
In-Reply-To: <CAGfcS_nQW4XhHsqHBPr_LtBDDKNhpvSFPNLAxtbRyHaUpLGaNA@mail.gmail.com>



>-----Original Message-----
>From: Rich Freeman <rich0@gentoo.org> 
>Sent: Monday, August 15, 2022 12:52 PM
>To: gentoo-user@lists.gentoo.org
>Subject: Re: [gentoo-user] Backup program that compresses data but only changes new files.
>
>On Mon, Aug 15, 2022 at 3:41 PM Dale <rdalek1967@gmail.com> wrote:
>>
>> Glad to know what I found was good info.  I just wonder how long it 
>> will be before even 10TB drives will be SMR.  I also dread having to 
>> search out a 14TB drive later.  :/
>>
>
>I think it will be a long time if ever, and here is why.
>
>There are good reasons and bad reasons to use SMR.  The reason you would WANT to use SMR is that you have a task that is well-suited to their limitations like backup or applications that can use log-style storage.  Ideally you'd want host-managed SMR for this.  The benefit is higher density for the cost, so you'd be doing it to get a drive that is cheaper than it otherwise would be.  However, these are all things that would appeal to experts who really know what they're doing.
>
>The bad reason to use SMR is that you're a manufacturer trying to squeeze out a bit more profit margin, not passing on the savings.  In this case you want to sell the drive to somebody who DOESN'T know what they're doing, and make it drive-managed.
>
>This is why we've seen SMR in medium-sized drives and not big ones as would be expected if you assumed it would be employed for the good reasons.  The only people buying 14TB hard drives are people who tend to know what they're doing, which makes them less of a target for unscrupulous manufacturers.  You wouldn't see them as much in small drives as the return in capacity isn't as much.  The medium sized drives are big enough to get a return out of using SMR, but small enough that suckers will be willing to buy them.
>
>At least, that's my theory...
>
>--
>Rich
>
>

A big chunk of it is that, when SMR drives came out, there was no reliable OS support for it, so it basically had to be drive-managed.  Which then had horrible performance, and the cherry on top was that the drive manufacturers tried to cover up what they'd changed.  So that made lots of the big companies doing big storage applications decide that SMR was crap and they simply will not buy SMR drives at any price.

Which is a real pity because there are lots of large-data applications where the write order is pretty much entirely sequential, so a properly designed, host managed system would see virtually no performance loss from SMR, and be able to take advantage of the higher density.

The moral is:  be transparent with your customers.

LMP

  parent reply	other threads:[~2022-08-15 21:11 UTC|newest]

Thread overview: 49+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-08-14 22:44 [gentoo-user] Backup program that compresses data but only changes new files Dale
2022-08-14 22:58 ` Mark Knecht
2022-08-14 23:20   ` Dale
2022-08-15  4:18     ` Wols Lists
2022-08-15 17:12     ` Mark Knecht
2022-08-15 19:19       ` Dale
2022-08-15 19:47         ` Mark Knecht
2022-08-15 20:17           ` Dale
2022-08-15 20:23             ` Mark Knecht
2022-08-16  5:25               ` Dale
2022-08-14 23:03 ` Rich Freeman
2022-08-15  7:02   ` John Covici
2022-08-15 10:50     ` Gerrit Kühn
2022-08-15 10:58       ` Gerrit Kühn
2022-08-15 11:43         ` Michael
2022-08-15 19:28           ` Dale
2022-08-15 19:43             ` Rich Freeman
2022-08-16  5:49           ` Gerrit Kühn
2022-08-15  7:05   ` Dale
2022-08-15 10:11     ` Rich Freeman
2022-08-15 18:32       ` Wol
2022-08-15 18:54         ` Rich Freeman
2022-08-15 18:40       ` J. Roeleveld
2022-08-15 17:24     ` Laurence Perkins
2022-08-15 18:34     ` J. Roeleveld
2022-08-15 18:56       ` Rich Freeman
2022-08-15 19:20         ` J. Roeleveld
2022-08-15 19:41           ` Rich Freeman
2022-08-14 23:08 ` Rich Freeman
2022-08-15  1:42 ` Julien Roy
2022-08-15  7:52   ` Dale
2022-08-15 18:47     ` Wol
2022-08-15 19:41       ` Dale
2022-08-15 19:51         ` Rich Freeman
2022-08-15 20:28           ` Dale
2022-08-15 21:11           ` Laurence Perkins [this message]
2022-08-16  5:17             ` Dale
2022-08-15 18:49     ` J. Roeleveld
2022-08-15 16:27   ` Matt Connell
2022-08-15  6:00 ` William Kenworthy
2022-08-15  8:33   ` Dale
2022-08-15  9:45     ` John Covici
2022-08-15 18:15       ` Wol
2022-08-15 19:46         ` Dale
2022-08-15 21:03           ` Laurence Perkins
2022-08-15  6:39 ` J. Roeleveld
2022-08-15 19:07   ` Dale
2022-08-15 19:25     ` J. Roeleveld
2022-08-16  5:33       ` Dale

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=MW2PR07MB405832F3410A6FFA2C67BA5BD2689@MW2PR07MB4058.namprd07.prod.outlook.com \
    --to=lperkins@openeye.net \
    --cc=gentoo-user@lists.gentoo.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox