bristol@mailman.lug.org.uk
To subscribe or unsubscribe via the World Wide Web, visit
https://mailman.lug.org.uk/mailman/listinfo/bristol
or, via email, send a message with subject or body 'help' to
bristol-request@mailman.lug.org.uk
You can reach the person managing the list at
bristol-owner@mailman.lug.org.uk
When replying, please edit your Subject line so it is more specific
than "Re: Contents of Bristol digest..."
Today's Topics:
1. Re: Postgres optimised h/w (Alex Butcher)
2. Re: Postgres optimised h/w (Alex Butcher)
3. Re: Postgres optimised h/w (Alex Butcher)
4. Re: Postgres optimised h/w (Alex Butcher)
----------------------------------------------------------------------
Message: 1
Date: Sat, 17 Dec 2016 11:36:33 +0000 (GMT)
From: Alex Butcher <lug@assursys.co.uk>
To: Martin Moore <martinm@it-helps.co.uk>, Bristol and Bath Linux
User Group <bristol@mailman.lug.org.uk>
Cc: Martin <inkubus@interalpha.co.uk>
Subject: Re: [bristol] Postgres optimised h/w
Message-ID:
<alpine.LRH.2.11.1612171134340.753@zlgugi.of5.nffheflf.cev>
Content-Type: text/plain; charset="utf-8"; Format="flowed"
On Fri, 16 Dec 2016, Martin Moore via Bristol wrote:
> I've also heard that h/w RAID controllers use their own format which
> renders them useless without the controller.
Not so much their own format, as metadata at one end or the other of the
member discs so that arrays can be automatically assembled/rebuilt. This
does mean that tools such as fdisk may not be able to find the partition
table, because they don't know to skip the metadata.
Best Regards,
Alex
------------------------------
Message: 2
Date: Sat, 17 Dec 2016 11:41:04 +0000 (GMT)
From: Alex Butcher <lug@assursys.co.uk>
To: debian@invux.com, Bristol and Bath Linux User Group
<bristol@mailman.lug.org.uk>
Subject: Re: [bristol] Postgres optimised h/w
Message-ID:
<alpine.LRH.2.11.1612171136420.753@zlgugi.of5.nffheflf.cev>
Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed
On Fri, 16 Dec 2016, Steve King via Bristol wrote:
> Plus you have to replace Cache RAM batteries.
Not necessarily; Adaptec's 5Z series of controllers from 2009 include
persistent flash and a supercapacitor to give the controller enough power to
flush write buffers to it in the event of power failure.
<http://www.theregister.co.uk/2009/06/25/adaptec_capacitor/>.
Also, to compare apples with apples, you have to also replace UPS batteries.
> I would recommend ECC RAM though.
Seconded, for any server application. Heck, I almost went with ECC on my
last home desktop after observing a few bitflips when copying 10s of GB
across my LAN. My wallet said no, however.
> Steve
Best Regards,
Alex
------------------------------
Message: 3
Date: Sat, 17 Dec 2016 11:49:44 +0000 (GMT)
From: Alex Butcher <lug@assursys.co.uk>
To: Amias Channer <me@amias.net>, Bristol and Bath Linux User Group
<bristol@mailman.lug.org.uk>
Cc: Martin Moore <martinm@it-helps.co.uk>, Martin
<inkubus@interalpha.co.uk>
Subject: Re: [bristol] Postgres optimised h/w
Message-ID:
<alpine.LRH.2.11.1612171141220.753@zlgugi.of5.nffheflf.cev>
Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed
On Fri, 16 Dec 2016, Amias Channer via Bristol wrote:
> I would definitely agree that software raid is as good as hardware
> raid unless you are a mega corporation, in which case just do it all
> in live RAM anyway. The complexity and proprietary formats make them
> dangerous.
See my point in a previous message about operational costs. It's easy as a
nerd to believe that /anyone/ can successfully use mdadm to rebuild a
degraded array once a disc has failed, but I'd not put money on it in many
organisations.
> Server bioses don't need any more reasons to boot up slowly
> which is usually the case with hardware raid.
Meh. How often do you reboot physical servers?
> Server storage is IMHO really about concurrent access and RAID helped
> improve that on hard disks but its pretty much pointless on SSD/NVMe
> solutions where nothing is spinning and an individual drive can
> saturate the controller. Mirroring in this context is useful but
> really only to slower storage like HDD.
A high performance database server should be aiming to saturate the PCIe
bus (15.75GB/s for PCIe 3.0 x16, 31.51GB/s for PCIe 4.0 x16). That might
require an array of striped (and mirrored, for availability!) SSDs.
One thing a hardware RAID controller does complicate is aligning
partitions/LVM physical/logical volumes and filesystem data structures with
the stripe size, so as to ensure the maximum throughput with multiple
simultaneous threads. Linux's md RAID layer passes up the stripe size of
arrays to higher layers so they can align themselves automagically, but with
hardware RAID, that's the sysadmin's job, sadly.
<https://www.percona.com/blog/2011/06/09/aligning-io-on-a-hard-disk-raid-the-theory/>
<https://anosi.blogspot.co.uk/2009/07/lvm-installation-partition-alignment.html>
<https://web.archive.org/web/20150907062944/http://www.penguincomputing.com/current-thinking-regarding-xfs-and-raid-arrays/>
<http://blog.tsunanet.net/2011/08/mkfsxfs-raid10-optimal-performance.html>
> Cheers
> Amias
Best Regards,
Alex
------------------------------
Message: 4
Date: Sat, 17 Dec 2016 11:54:33 +0000 (GMT)
From: Alex Butcher <lug@assursys.co.uk>
To: Nigel Sollars <nsollars@gmail.com>, Bristol and Bath Linux User
Group <bristol@mailman.lug.org.uk>
Cc: Amias Channer <me@amias.net>, Martin <inkubus@interalpha.co.uk>,
Martin Moore <martinm@it-helps.co.uk>
Subject: Re: [bristol] Postgres optimised h/w
Message-ID:
<alpine.LRH.2.11.1612171150020.753@zlgugi.of5.nffheflf.cev>
Content-Type: text/plain; charset="utf-8"; Format="flowed"
On Fri, 16 Dec 2016, Nigel Sollars via Bristol wrote:
> Isnt raid normally for Corps anyhow?... I mean really unless your a data
> hungry monster whats the point?..
Naw, I've been using it on my home desktops since 2002. My first 120MB HDD
and SCSI controller was nearly £500 in 1993 money, so getting 2x4TB HDDs for
~£270 of relatively-less-valuable-2016 money (about £147 of 1993 money,
according to
<http://www.bankofengland.co.uk/education/Pages/resources/inflationtools/calculator/flash/default.aspx>)
makes RAID a no-brainer for me.
Also, many modern laptops use Intel's Smart Reponse technology which
effectively builds a partial RAID 1 array between an SSD and a HDD to give
the best of both worlds.
Best Regards,
Alex
------------------------------
Subject: Digest Footer
_______________________________________________
Bristol mailing list
Bristol@mailman.lug.org.uk
https://mailman.lug.org.uk/mailman/listinfo/bristol
------------------------------
End of Bristol Digest, Vol 673, Issue 11
****************************************
Tidak ada komentar:
Posting Komentar