Re: AIX - JFS2 Snapshots
From: Mladen Gogala <gogala.mladen_at_gmail.com>
Date: Tue, 23 Feb 2016 10:35:00 -0500
Message-ID: <56CC7C24.9000108_at_gmail.com>
On 02/23/2016 09:33 AM, Ludovico Caldara wrote:
> Hi
>
> _at_Mladen, it's not the case anymore. Most recent snapshot technologies
> do the so-called "redirect on write" (or "copy on first write")
> mechanism that limits the necessity of additional I/O, at the expense
> of higher file fragmentation. ACFS is an example easy to verify with
> `acfsutil info file`.
>
> _at_Jeremy: beware that if you use TTS in conjunction with snapshots, you
> will end up by having inter-dependencies between 1 LUN and many
> Databases, so plan carefully your strategy, in particular naming
> conventions and snapshots and LUNs creation and deletion.
>
> just my 2 cents...
Hi Lodovico,
We're discussing JFS2 which is a good decade and a half older than ACFS. It's even older than OCFS, in case someone still remembers it. Second, "redirect on write" or "deferred write", is used by most SAN vendors. That is why I made the distinction between SAN snapshot and JFS2 snapshot. Third, low end file systems like BRTFS still use COW, despite being a good amount of time newer than JFS2. Fact of the matter is that Linux LVM and most file systems, like NTFS, BRTFS and ZFS are using COW, because it's the easiest one to implement and doesn't require space for growth. With "redirect on write", block are much larger, typically 32 or 64MB and your snapshot will not kill your system but will consume space at much faster pace.
Don't get me wrong, snapshots are great tools, as long as you are aware what they are and how to use them.
Date: Tue, 23 Feb 2016 10:35:00 -0500
Message-ID: <56CC7C24.9000108_at_gmail.com>
On 02/23/2016 09:33 AM, Ludovico Caldara wrote:
> Hi
>
> _at_Mladen, it's not the case anymore. Most recent snapshot technologies
> do the so-called "redirect on write" (or "copy on first write")
> mechanism that limits the necessity of additional I/O, at the expense
> of higher file fragmentation. ACFS is an example easy to verify with
> `acfsutil info file`.
>
> _at_Jeremy: beware that if you use TTS in conjunction with snapshots, you
> will end up by having inter-dependencies between 1 LUN and many
> Databases, so plan carefully your strategy, in particular naming
> conventions and snapshots and LUNs creation and deletion.
>
> just my 2 cents...
Hi Lodovico,
We're discussing JFS2 which is a good decade and a half older than ACFS. It's even older than OCFS, in case someone still remembers it. Second, "redirect on write" or "deferred write", is used by most SAN vendors. That is why I made the distinction between SAN snapshot and JFS2 snapshot. Third, low end file systems like BRTFS still use COW, despite being a good amount of time newer than JFS2. Fact of the matter is that Linux LVM and most file systems, like NTFS, BRTFS and ZFS are using COW, because it's the easiest one to implement and doesn't require space for growth. With "redirect on write", block are much larger, typically 32 or 64MB and your snapshot will not kill your system but will consume space at much faster pace.
Don't get me wrong, snapshots are great tools, as long as you are aware what they are and how to use them.
-- Mladen Gogala Oracle DBA http://mgogala.freehostia.com -- http://www.freelists.org/webpage/oracle-lReceived on Tue Feb 23 2016 - 16:35:00 CET