Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > uk.comp.sys.mac > #117443 > unrolled thread
| Started by | glawrie <gavin.lawrie@2gc.eu> |
|---|---|
| First post | 2017-04-26 19:20 +0100 |
| Last post | 2017-05-01 09:02 +0000 |
| Articles | 20 on this page of 46 — 12 participants |
Back to article view | Back to uk.comp.sys.mac
Five years later... Intel launches Fusion Drive for Wintel... glawrie <gavin.lawrie@2gc.eu> - 2017-04-26 19:20 +0100
Re: Five years later... Intel launches Fusion Drive for Wintel... Theo <theom+news@chiark.greenend.org.uk> - 2017-04-26 19:47 +0100
Re: Five years later... Intel launches Fusion Drive for Wintel... JF Mezei <jfmezei.spamnot@vaxination.ca> - 2017-04-26 16:35 -0400
Re: Five years later... Intel launches Fusion Drive for Wintel... Ian McCall <ian@eruvia.org> - 2017-04-27 08:04 +0100
Re: Five years later... Intel launches Fusion Drive for Wintel... Lewis <g.kreme@gmail.com.dontsendmecopies> - 2017-04-27 09:41 +0000
Re: Five years later... Intel launches Fusion Drive for Wintel... JF Mezei <jfmezei.spamnot@vaxination.ca> - 2017-04-27 13:52 -0400
Re: Five years later... Intel launches Fusion Drive for Wintel... Jolly Roger <jollyroger@pobox.com> - 2017-04-27 18:19 +0000
Re: Five years later... Intel launches Fusion Drive for Wintel... nmassello@yahoo.com (Neill Massello) - 2017-04-27 20:54 -0600
Re: Five years later... Intel launches Fusion Drive for Wintel... Jolly Roger <jollyroger@pobox.com> - 2017-04-28 05:30 +0000
Re: Five years later... Intel launches Fusion Drive for Wintel... Lewis <g.kreme@gmail.com.dontsendmecopies> - 2017-04-28 11:14 +0000
Re: Five years later... Intel launches Fusion Drive for Wintel... JF Mezei <jfmezei.spamnot@vaxination.ca> - 2017-04-28 14:49 -0400
Re: Five years later... Intel launches Fusion Drive for Wintel... Jolly Roger <jollyroger@pobox.com> - 2017-04-28 19:08 +0000
Re: Five years later... Intel launches Fusion Drive for Wintel... JF Mezei <jfmezei.spamnot@vaxination.ca> - 2017-04-28 22:56 -0400
Re: Five years later... Intel launches Fusion Drive for Wintel... Jolly Roger <jollyroger@pobox.com> - 2017-04-29 14:34 +0000
Re: Five years later... Intel launches Fusion Drive for Wintel... Lewis <g.kreme@gmail.com.dontsendmecopies> - 2017-04-29 15:19 +0000
Re: Five years later... Intel launches Fusion Drive for Wintel... Alan Baker <alangbaker@telus.net> - 2017-04-28 21:30 -0700
Re: Five years later... Intel launches Fusion Drive for Wintel... Elliott Roper <nospam@yrl.co.uk> - 2017-04-29 15:18 +1000
Re: Five years later... Intel launches Fusion Drive for Wintel... Jaimie Vandenbergh <jaimie@sometimes.sessile.org> - 2017-04-29 09:45 +0100
Re: Five years later... Intel launches Fusion Drive for Wintel... nospam <nospam@nospam.invalid> - 2017-04-29 09:30 -0400
Re: Five years later... Intel launches Fusion Drive for Wintel... Jaimie Vandenbergh <jaimie@sometimes.sessile.org> - 2017-04-29 14:58 +0100
Re: Five years later... Intel launches Fusion Drive for Wintel... nospam <nospam@nospam.invalid> - 2017-04-29 10:22 -0400
Re: Five years later... Intel launches Fusion Drive for Wintel... JF Mezei <jfmezei.spamnot@vaxination.ca> - 2017-04-29 14:13 -0400
Re: Five years later... Intel launches Fusion Drive for Wintel... nospam <nospam@nospam.invalid> - 2017-04-29 14:19 -0400
Re: Five years later... Intel launches Fusion Drive for Wintel... JF Mezei <jfmezei.spamnot@vaxination.ca> - 2017-04-29 14:52 -0400
Re: Five years later... Intel launches Fusion Drive for Wintel... nospam <nospam@nospam.invalid> - 2017-04-29 14:57 -0400
Re: Five years later... Intel launches Fusion Drive for Wintel... JF Mezei <jfmezei.spamnot@vaxination.ca> - 2017-04-29 15:32 -0400
Re: Five years later... Intel launches Fusion Drive for Wintel... nospam <nospam@nospam.invalid> - 2017-04-29 15:32 -0400
Re: Five years later... Intel launches Fusion Drive for Wintel... Lewis <g.kreme@gmail.com.dontsendmecopies> - 2017-04-29 19:53 +0000
Re: Five years later... Intel launches Fusion Drive for Wintel... Ian McCall <ian@eruvia.org> - 2017-04-29 22:59 +0100
Re: Five years later... Intel launches Fusion Drive for Wintel... Jolly Roger <jollyroger@pobox.com> - 2017-04-29 22:04 +0000
Re: Five years later... Intel launches Fusion Drive for Wintel... JF Mezei <jfmezei.spamnot@vaxination.ca> - 2017-04-29 14:07 -0400
Re: Five years later... Intel launches Fusion Drive for Wintel... Lewis <g.kreme@gmail.com.dontsendmecopies> - 2017-04-29 20:54 +0000
Re: Five years later... Intel launches Fusion Drive for Wintel... Jolly Roger <jollyroger@pobox.com> - 2017-04-29 14:44 +0000
Re: Five years later... Intel launches Fusion Drive for Wintel... JF Mezei <jfmezei.spamnot@vaxination.ca> - 2017-04-29 01:35 -0400
Re: Five years later... Intel launches Fusion Drive for Wintel... Jaimie Vandenbergh <jaimie@sometimes.sessile.org> - 2017-04-29 09:39 +0100
Re: Five years later... Intel launches Fusion Drive for Wintel... Jolly Roger <jollyroger@pobox.com> - 2017-04-29 14:49 +0000
Re: Five years later... Intel launches Fusion Drive for Wintel... Lewis <g.kreme@gmail.com.dontsendmecopies> - 2017-04-29 19:58 +0000
Re: Five years later... Intel launches Fusion Drive for Wintel... nmassello@yahoo.com (Neill Massello) - 2017-04-29 16:08 -0600
Re: Five years later... Intel launches Fusion Drive for Wintel... Fred <fredb@[127.0.0.1]> - 2017-04-29 23:35 +0100
Re: Five years later... Intel launches Fusion Drive for Wintel... Jolly Roger <jollyroger@pobox.com> - 2017-04-29 22:44 +0000
Re: Five years later... Intel launches Fusion Drive for Wintel... Alan Baker <alangbaker@telus.net> - 2017-04-30 22:45 -0700
Re: Five years later... Intel launches Fusion Drive for Wintel... JF Mezei <jfmezei.spamnot@vaxination.ca> - 2017-05-01 02:09 -0400
Re: Five years later... Intel launches Fusion Drive for Wintel... Alan Baker <alangbaker@telus.net> - 2017-04-30 23:20 -0700
Re: Five years later... Intel launches Fusion Drive for Wintel... Ian McCall <ian@eruvia.org> - 2017-05-01 09:28 +0100
Re: Five years later... Intel launches Fusion Drive for Wintel... JF Mezei <jfmezei.spamnot@vaxination.ca> - 2017-05-01 13:30 -0400
Re: Five years later... Intel launches Fusion Drive for Wintel... Lewis <g.kreme@gmail.com.dontsendmecopies> - 2017-05-01 09:02 +0000
Page 1 of 3 [1] 2 3 Next page →
| From | glawrie <gavin.lawrie@2gc.eu> |
|---|---|
| Date | 2017-04-26 19:20 +0100 |
| Subject | Five years later... Intel launches Fusion Drive for Wintel... |
| Message-ID | <2017042619201161432-gavin.lawrie@2gc.eu> |
https://hardware.slashdot.org/story/17/04/25/0617219/intel-launches-optane-memory-that-makes-standard-hard-drives-perform-like-ssds Apple introduced Fusion Drives in 2012. It combines an SSD with a spinning disk to give appearance of a larger single disk; software on the host computer manages the files stored on the SSD to try and keep most-used files on the SSD and boost performance of the computer. The Intel version appears to do the same thing. It features some very fast fancy memory tech, but it only works on a handful of systems (currently its use is limited to "Windows 10 64-bit systems with Intel 7th Gen Kaby Lake-based processors and 200-series chipsets, or newer"). Once again looks like Apple were a few years ahead of the curve... ;)
[toc] | [next] | [standalone]
| From | Theo <theom+news@chiark.greenend.org.uk> |
|---|---|
| Date | 2017-04-26 19:47 +0100 |
| Message-ID | <dMf*mRdqw@news.chiark.greenend.org.uk> |
| In reply to | #117443 |
In uk.comp.sys.mac glawrie <gavin.lawrie@2gc.eu> wrote: > https://hardware.slashdot.org/story/17/04/25/0617219/intel-launches-optane-memory-that-makes-standard-hard-drives-perform-like-ssds > > > Apple introduced Fusion Drives in 2012. It combines an SSD with a > spinning disk to give appearance of a larger single disk; software on > the host computer manages the files stored on the SSD to try and keep > most-used files on the SSD and boost performance of the computer. > > The Intel version appears to do the same thing. It features some very > fast fancy memory tech, but it only works on a handful of systems > (currently its use is limited to "Windows 10 64-bit systems with Intel > 7th Gen Kaby Lake-based processors and 200-series chipsets, or newer"). > > Once again looks like Apple were a few years ahead of the curve... ;) Intel have had 'Smart Response Technology' since 2011, which is pretty much the same thing. Unfortunately they love market segmentation, and the segment that would have benefited from it (that couldn't afford large SSDs) wasn't the one that could use it. It seems like they're playing the same game again. While Fusion Drive/SRT is a bit of a waste of time with Optane (just use an SSD as cache instead), I can think of a few uses of 32GB of fast 'swap'. It would be fun to buy an Optane drive and see if it'll work on non-Kaby Lake machines. I can't imagine why not - it's just NVMe after all. Though maybe they've hobbled them via the BIOS. Theo
[toc] | [prev] | [next] | [standalone]
| From | JF Mezei <jfmezei.spamnot@vaxination.ca> |
|---|---|
| Date | 2017-04-26 16:35 -0400 |
| Message-ID | <590104a9$0$36059$c3e8da3$76a7c58f@news.astraweb.com> |
| In reply to | #117443 |
On 2017-04-26 14:20, glawrie wrote: > https://hardware.slashdot.org/story/17/04/25/0617219/intel-launches-optane-memory-that-makes-standard-hard-drives-perform-like-ssds > The Intel version appears to do the same thing. Very very different. Apple is glorified Raid 1 with software to automatically and dynamically manage placement of FILES on the multi-disk logical volume. And it does increase storage camapcity. Intel's is a hardware based solution which operates at the disk BLOCK level, not at file system level. It merely cases onto SSD more frequently used disk blocks (irrespective of what file they belong to since hardware is not aware of a file system). Would need to know how Intel handles writes in terms of when a write is garanteed to have been made. (think power failure between time the write has been made on SSD and confirmed back to OS and time the cached-on-SSD version makes it to the spinning drive, and whether when powered back up, the hardeware remembers the transfers that still need to be done between SSD and spinning drive.) With disk block virtualization, it becomes possible for Intel's solution to theoretically extend storage by mapping most frequently used virtual blocks onto disk blocks on SSD (which are them selves mapped onto actual SSD physical location) or on the spinning drive. This increases storage AND removes the issue of cache coherence since the SSD isn't a cache anymore. So the details of implementation matter. Intel would be a hardware based Raid 1 solution it it cobines disk blocks from multiple disks (as opposed to being a glorified cache).
[toc] | [prev] | [next] | [standalone]
| From | Ian McCall <ian@eruvia.org> |
|---|---|
| Date | 2017-04-27 08:04 +0100 |
| Message-ID | <emdjeoF5f0aU1@mid.individual.net> |
| In reply to | #117447 |
On 2017-04-26 20:35:52 +0000, JF Mezei <jfmezei.spamnot@vaxination.ca> said: > On 2017-04-26 14:20, glawrie wrote: >> https://hardware.slashdot.org/story/17/04/25/0617219/intel-launches-optane-memory-that-makes-standard-hard-drives-perform-like-ssds The >> >> Intel version appears to do the same thing. > > Very very different. Apple is glorified Raid 1 with software to > automatically and dynamically manage placement of FILES on the > multi-disk logical volume. And it does increase storage camapcity. > > Intel's is a hardware based solution which operates at the disk BLOCK > level, not at file system level. It merely cases onto SSD more > frequently used disk blocks (irrespective of what file they belong to > since hardware is not aware of a file system). Small clarification because I think I know what you're getting at: Fusion works at block level, not at file level. However it does require a certain file system in order to do so, so it's work at the file system's idea of a block, not necessarily the hardware's idea of a block. Cheers, Ian
[toc] | [prev] | [next] | [standalone]
| From | Lewis <g.kreme@gmail.com.dontsendmecopies> |
|---|---|
| Date | 2017-04-27 09:41 +0000 |
| Message-ID | <slrnog3fbe.jnv.g.kreme@snow.local> |
| In reply to | #117447 |
In message <590104a9$0$36059$c3e8da3$76a7c58f@news.astraweb.com> JF Mezei <jfmezei.spamnot@vaxination.ca> wrote: > On 2017-04-26 14:20, glawrie wrote: >> https://hardware.slashdot.org/story/17/04/25/0617219/intel-launches-optane-memory-that-makes-standard-hard-drives-perform-like-ssds >> The Intel version appears to do the same thing. > Very very different. Apple is glorified Raid 1 with software to > automatically and dynamically manage placement of FILES on the > multi-disk logical volume. And it does increase storage camapcity. No, it is nothing at all like RAID 1. Nothing. At all. Fusion creates a single core storage volume using up the combined space of the HDD and the SSD. Ther is no mirroring involved. Frequently accessed files are managed by CoreStorage and moved to the SSD portion of the fusion drive without the knowledge or intervention of the OS or the applications. Most used files are simply on a faster medium. Fusion worked very well, but I am sure Intel's new version, being restricted to the newest OS, newest chips, and therefore newest machines, will perform better. Nothing beats a 500GB or larger SSD boot drive though, and Apple has moved on from spinning hard drives. Even the bottom of the line 2015 Mac mini has speed benchmarks in excess of 500MB/s read/write. -- ...but the senator, while insisting he was not intoxicated, could not explain his nudity.
[toc] | [prev] | [next] | [standalone]
| From | JF Mezei <jfmezei.spamnot@vaxination.ca> |
|---|---|
| Date | 2017-04-27 13:52 -0400 |
| Message-ID | <59022ff3$0$22780$c3e8da3$33881b6a@news.astraweb.com> |
| In reply to | #117456 |
On 2017-04-27 05:41, Lewis wrote: > No, it is nothing at all like RAID 1. Nothing. At all. Sorry, Raid 0. Often forget the numbers for bound sets and mirroring. Fusion is bound volume so yeah, Raid 0. > Fusion worked very well, but I am sure Intel's new version, being > restricted to the newest OS, newest chips, and therefore newest > machines, will perform better. Is Intel's really tied to an OS? From the reading, it seems to present a normal disk drive to the OS. But would need appropriate hardware support so the 2 disks are joined and one disk presented to OS at boot time. (all the way down to EFI support) > Nothing beats a 500GB or larger SSD boot drive though, In the desktop area, as SSDs of sufficient size before affordable, yeah. And once all your storage is SSD, the need to move files to certain blocks for faster access becomes moot. But in the enterprise area, disk arrays have, for a long time done what Intel is doing by caching data in the array's RAM to provide better response time. (and this has implications for transactional since the disk responds to OS with "write done" when it is stored in RAM, not necessarily stored on disk (so UPS is absolutely required or sufficient autonomy to complete all writes to spinning disks after a power failure)
[toc] | [prev] | [next] | [standalone]
| From | Jolly Roger <jollyroger@pobox.com> |
|---|---|
| Date | 2017-04-27 18:19 +0000 |
| Message-ID | <emer0vFd5veU2@mid.individual.net> |
| In reply to | #117465 |
On 2017-04-27, JF Mezei <jfmezei.spamnot@vaxination.ca> wrote: > On 2017-04-27 05:41, Lewis wrote: > >> No, it is nothing at all like RAID 1. Nothing. At all. > Sorry, Raid 0. > > Often forget the numbers for bound sets and mirroring. Fusion is bound > volume so yeah, Raid 0. Nope. You really have no idea what Core Storage is about, do you? -- E-mail sent to this address may be devoured by my ravenous SPAM filter. I often ignore posts from Google. Use a real news client instead. JR
[toc] | [prev] | [next] | [standalone]
| From | nmassello@yahoo.com (Neill Massello) |
|---|---|
| Date | 2017-04-27 20:54 -0600 |
| Message-ID | <1n55ra2.9fnkb8ze8hw8N%nmassello@yahoo.com> |
| In reply to | #117465 |
JF Mezei <jfmezei.spamnot@vaxination.ca> wrote: > Often forget the numbers for bound sets and mirroring. Fusion is bound > volume so yeah, Raid 0. Because you're so fuzzy on all things RAID, you really shouldn't use the term. It's just asking for an argument. (RAID 0 is simple striping.)
[toc] | [prev] | [next] | [standalone]
| From | Jolly Roger <jollyroger@pobox.com> |
|---|---|
| Date | 2017-04-28 05:30 +0000 |
| Message-ID | <emg2chFk68vU1@mid.individual.net> |
| In reply to | #117473 |
Neill Massello <nmassello@yahoo.com> wrote: > JF Mezei <jfmezei.spamnot@vaxination.ca> wrote: > >> Often forget the numbers for bound sets and mirroring. Fusion is bound >> volume so yeah, Raid 0. > > Because you're so fuzzy on all things RAID, you really shouldn't use the > term. It's just asking for an argument. (RAID 0 is simple striping.) He's also clueless about Fusion Drives. He doesn't know when to keep his mouth shut. He'd rather spread misinformation and FUD than actually learn anything. It's his modus operandi. -- E-mail sent to this address may be devoured by my ravenous SPAM filter. I often ignore posts from Google. Use a real news client instead. JR
[toc] | [prev] | [next] | [standalone]
| From | Lewis <g.kreme@gmail.com.dontsendmecopies> |
|---|---|
| Date | 2017-04-28 11:14 +0000 |
| Message-ID | <slrnog696f.lhc.g.kreme@snow.local> |
| In reply to | #117465 |
In message <59022ff3$0$22780$c3e8da3$33881b6a@news.astraweb.com> JF Mezei <jfmezei.spamnot@vaxination.ca> wrote: > On 2017-04-27 05:41, Lewis wrote: >> No, it is nothing at all like RAID 1. Nothing. At all. > Sorry, Raid 0. It's also nothing like raid-0 which stripes files across multiple drives. Care to try again? > Often forget the numbers for bound sets and mirroring. Fusion is bound > volume so yeah, Raid 0. No. Not at all. -- Lady Astor: "If you were my husband I'd give you poison." Churchill: "If you were my wife, I'd drink it."
[toc] | [prev] | [next] | [standalone]
| From | JF Mezei <jfmezei.spamnot@vaxination.ca> |
|---|---|
| Date | 2017-04-28 14:49 -0400 |
| Message-ID | <59038ed4$0$10750$c3e8da3$5d8fb80f@news.astraweb.com> |
| In reply to | #117478 |
On 2017-04-28 07:14, Lewis wrote: > It's also nothing like raid-0 which stripes files across multiple > drives. Care to try again? Raid 0 does not necessarily stripe files across multiple volumes. VMS for instance couldn't do that because the file ID (the inode in Unix parlance) contained a set of bits to denote on which volume of the set the file resided on). Another OS may be able to stripe a single file across multiple drives. What is common to Raid 0 is a common file system (1 catalogue, 1 mount point in Unix parlance and you get a single directory of files irrespective of whcih physical disks they are on. Bound volume sets were mostly used in the days of 600MB disks (mid 1980s, so well before you were born) when you needed more storage on a single drive. (and when the OS didn't handle bigger disks due to the OS software lacking sufficient bits to address larger disks, so carving a logical disk into multiple physical ones solved that issue). Once larger disks (and OS supoort for very large number of blocks) arrived, disk striping stopped being used, until SSDs came in and it was again fashionable because of applications such as Fusion drives.
[toc] | [prev] | [next] | [standalone]
| From | Jolly Roger <jollyroger@pobox.com> |
|---|---|
| Date | 2017-04-28 19:08 +0000 |
| Message-ID | <emhi97Ftb4jU2@mid.individual.net> |
| In reply to | #117480 |
On 2017-04-28, JF Mezei <jfmezei.spamnot@vaxination.ca> wrote: > On 2017-04-28 07:14, Lewis wrote: > >> It's also nothing like raid-0 which stripes files across multiple >> drives. Care to try again? > > Raid 0 Core Storage-based Fusion Drives aren't in any way equivalent to RAID 0. -- E-mail sent to this address may be devoured by my ravenous SPAM filter. I often ignore posts from Google. Use a real news client instead. JR
[toc] | [prev] | [next] | [standalone]
| From | JF Mezei <jfmezei.spamnot@vaxination.ca> |
|---|---|
| Date | 2017-04-28 22:56 -0400 |
| Message-ID | <590400e6$0$34646$b1db1813$15bdbe48@news.astraweb.com> |
| In reply to | #117481 |
On 2017-04-28 15:08, Jolly Roger wrote: > Core Storage-based Fusion Drives aren't in any way equivalent to RAID 0. They are a superset of Raid 0. But the core principle is a bound volume set. It has more sophisticated software to place file on approroate volumes and dynamically move then when beneficial. But still Raid 0.
[toc] | [prev] | [next] | [standalone]
| From | Jolly Roger <jollyroger@pobox.com> |
|---|---|
| Date | 2017-04-29 14:34 +0000 |
| Message-ID | <emjmkcFbc1kU1@mid.individual.net> |
| In reply to | #117486 |
JF Mezei <jfmezei.spamnot@vaxination.ca> wrote: > On 2017-04-28 15:08, Jolly Roger wrote: > >> Core Storage-based Fusion Drives aren't in any way equivalent to RAID 0. > > > They are a superset of Raid 0. No. Core Storage does not use RAID 0 in any way. You just can't stop spewing misinformation from your big mouth, can you? -- E-mail sent to this address may be devoured by my ravenous SPAM filter. I often ignore posts from Google. Use a real news client instead. JR
[toc] | [prev] | [next] | [standalone]
| From | Lewis <g.kreme@gmail.com.dontsendmecopies> |
|---|---|
| Date | 2017-04-29 15:19 +0000 |
| Message-ID | <slrnog9bum.2suk.g.kreme@snow.local> |
| In reply to | #117486 |
In message <590400e6$0$34646$b1db1813$15bdbe48@news.astraweb.com> JF Mezei <jfmezei.spamnot@vaxination.ca> wrote: > On 2017-04-28 15:08, Jolly Roger wrote: >> Core Storage-based Fusion Drives aren't in any way equivalent to RAID 0. > They are a superset of Raid 0. But the core principle is a bound volume > set. You have no fucking idea what you are talking about. Again. -- I have a cunning plan.
[toc] | [prev] | [next] | [standalone]
| From | Alan Baker <alangbaker@telus.net> |
|---|---|
| Date | 2017-04-28 21:30 -0700 |
| Message-ID | <oe14u4$15r6$2@gioia.aioe.org> |
| In reply to | #117465 |
On 2017-04-27 10:52 AM, JF Mezei wrote: > On 2017-04-27 05:41, Lewis wrote: > >> No, it is nothing at all like RAID 1. Nothing. At all. > Sorry, Raid 0. It's nothing like RAID 0, either. > > Often forget the numbers for bound sets and mirroring. Fusion is bound > volume so yeah, Raid 0. No. Utterly wrong. > > >> Fusion worked very well, but I am sure Intel's new version, being >> restricted to the newest OS, newest chips, and therefore newest >> machines, will perform better. > > Is Intel's really tied to an OS? From the reading, it seems to present a > normal disk drive to the OS. But would need appropriate hardware support > so the 2 disks are joined and one disk presented to OS at boot time. > (all the way down to EFI support) > >> Nothing beats a 500GB or larger SSD boot drive though, > > In the desktop area, as SSDs of sufficient size before affordable, yeah. > > And once all your storage is SSD, the need to move files to certain > blocks for faster access becomes moot. > > > But in the enterprise area, disk arrays have, for a long time done what > Intel is doing by caching data in the array's RAM to provide better > response time. (and this has implications for transactional since the > disk responds to OS with "write done" when it is stored in RAM, not > necessarily stored on disk (so UPS is absolutely required or sufficient > autonomy to complete all writes to spinning disks after a power failure) > >
[toc] | [prev] | [next] | [standalone]
| From | Elliott Roper <nospam@yrl.co.uk> |
|---|---|
| Date | 2017-04-29 15:18 +1000 |
| Message-ID | <0001HW.1EB459B6006286347000046642CF@news.giganews.com> |
| In reply to | #117487 |
On 29 Apr 2017, Alan Baker wrote (in article <oe14u4$15r6$2@gioia.aioe.org>): > On 2017-04-27 10:52 AM, JF Mezei wrote: > > On 2017-04-27 05:41, Lewis wrote: > > > > > No, it is nothing at all like RAID 1. Nothing. At all. > > Sorry, Raid 0. > > It's nothing like RAID 0, either. > > > > > Often forget the numbers for bound sets and mirroring. Fusion is bound > > volume so yeah, Raid 0. > > No. Utterly wrong. > > > > > > > > Fusion worked very well, but I am sure Intel's new version, being > > > restricted to the newest OS, newest chips, and therefore newest > > > machines, will perform better. > > > > Is Intel's really tied to an OS? From the reading, it seems to present a > > normal disk drive to the OS. But would need appropriate hardware support > > so the 2 disks are joined and one disk presented to OS at boot time. > > (all the way down to EFI support) > > > > > Nothing beats a 500GB or larger SSD boot drive though, > > > > In the desktop area, as SSDs of sufficient size before affordable, yeah. > > > > And once all your storage is SSD, the need to move files to certain > > blocks for faster access becomes moot. > > > > > > But in the enterprise area, disk arrays have, for a long time done what > > Intel is doing by caching data in the array's RAM to provide better > > response time. (and this has implications for transactional since the > > disk responds to OS with "write done" when it is stored in RAM, not > > necessarily stored on disk (so UPS is absolutely required or sufficient > > autonomy to complete all writes to spinning disks after a power failure) Who cares about Fusion drives? The idea is almost obsolete. If Intel get their 3D Xpoint Optane stuff affordable, it will be back to the drawing board for storage technology. Huge byte addressable non-volatile storage at the geometric mean of price and performance between DRAM and today's NVRAM. Spinning rust might end up in a niche with magtape, so slow that only museums will be interested. PS Hiya JF! Long time no see. -- To de-mung my e-mail address:- fsnospam$elliott$$ PGP Fingerprint: 1A96 3CF7 637F 896B C810 E199 7E5C A9E4 8E59 E248
[toc] | [prev] | [next] | [standalone]
| From | Jaimie Vandenbergh <jaimie@sometimes.sessile.org> |
|---|---|
| Date | 2017-04-29 09:45 +0100 |
| Message-ID | <6ck8gc5v8o957glg1d4hob8f2ian0nhtaq@4ax.com> |
| In reply to | #117488 |
On Sat, 29 Apr 2017 15:18:46 +1000, Elliott Roper <nospam@yrl.co.uk>
wrote:
>Who cares about Fusion drives? The idea is almost obsolete.
Was my first response when Apple released them. "That's neat but is a
very short term stopgap solution... SSDs will get bigger quickly, and
Apple will ditch spinning rust as soon as possible".
I was a little optimistic on both time and retirement of HDDs, but two
years later I put a 1Tb SSD in my iMac replacing the spinner for a
non-ridiculous price.
Cheers - Jaimie
--
"I'd tried caffeine a few times; it made me believe I was focused and
energetic, but it turned my judgment to shit. Widespread use of
caffeine explains a lot about the twentieth century."
- "Distress", Greg Egan
[toc] | [prev] | [next] | [standalone]
| From | nospam <nospam@nospam.invalid> |
|---|---|
| Date | 2017-04-29 09:30 -0400 |
| Message-ID | <290420170930404051%nospam@nospam.invalid> |
| In reply to | #117491 |
In article <6ck8gc5v8o957glg1d4hob8f2ian0nhtaq@4ax.com>, Jaimie Vandenbergh <jaimie@sometimes.sessile.org> wrote: > > >Who cares about Fusion drives? The idea is almost obsolete. > > Was my first response when Apple released them. "That's neat but is a > very short term stopgap solution... SSDs will get bigger quickly, and > Apple will ditch spinning rust as soon as possible". hds also continue to get bigger, and will always be cheaper per gigabyte, at least for the foreseeable future. > I was a little optimistic on both time and retirement of HDDs, but two > years later I put a 1Tb SSD in my iMac replacing the spinner for a > non-ridiculous price. and for roughly the same price, you could get a 6 terabyte hd.
[toc] | [prev] | [next] | [standalone]
| From | Jaimie Vandenbergh <jaimie@sometimes.sessile.org> |
|---|---|
| Date | 2017-04-29 14:58 +0100 |
| Message-ID | <7g69gch7rio045rnv4pme5dc8291jssj37@4ax.com> |
| In reply to | #117492 |
On Sat, 29 Apr 2017 09:30:40 -0400, nospam <nospam@nospam.invalid> wrote: >In article <6ck8gc5v8o957glg1d4hob8f2ian0nhtaq@4ax.com>, Jaimie >Vandenbergh <jaimie@sometimes.sessile.org> wrote: > >> >> >Who cares about Fusion drives? The idea is almost obsolete. >> >> Was my first response when Apple released them. "That's neat but is a >> very short term stopgap solution... SSDs will get bigger quickly, and >> Apple will ditch spinning rust as soon as possible". > >hds also continue to get bigger, and will always be cheaper per >gigabyte, at least for the foreseeable future. They'll become ever more niche over the next five years. Datacentre bulk storage only after that, where the economics will work for maybe another five years or so. But you can get 60TB SSD (not a typo) in a 3.5" size case. The writing is very much on the wall for spinning disks since they already only have price/byte advantage and that's temporary. Data density, speed, iops, watts per TB - SSDs already win. >> I was a little optimistic on both time and retirement of HDDs, but two >> years later I put a 1Tb SSD in my iMac replacing the spinner for a >> non-ridiculous price. > >and for roughly the same price, you could get a 6 terabyte hd. Yep. Which would be slow and noisy, not my aim. You could have worked that out with a little thought. I am not keen on shingled drives for general read/write usage either. Read-once-mostly, fine, but I wouldn't want one in my iMac. Cheers - Jaimie -- Many people in this group spent their school years taking illogical, pointless orders from morons and having their will to live systematically crushed. And people say school doesn't prepare kids for the real world. -- Rayner, asr
[toc] | [prev] | [next] | [standalone]
Page 1 of 3 [1] 2 3 Next page →
Back to top | Article view | uk.comp.sys.mac
csiph-web