Re: Raid

Tech Tip: Click here to run a free scan for Windows Errors and optimize PC performance

From: root (postmaster_at_buchanangc.com)
Date: 06/18/04 

Date: Thu, 17 Jun 2004 19:01:54 -0700

"Steve Foster [SBS MVP]" <steve.foster@picamar.co.uk> wrote in message
news:xn0djnkyp78s93700d@msnews.microsoft.com...
> root wrote:
>
> >
> > "Steve Foster [SBS MVP]" <steve.foster@picamar.co.uk> wrote in message
> > news:xn0djn1am6zcjnq006@msnews.microsoft.com...
> > > root wrote:
> > >
> > > > > The rest you would place according to performance requirements
> > > > > remembering that the mirror set will give you faster performance
> > > > > for reading
> > > >
> > > > No, RAID 5 is faster on reads.
> > >
> > > You've stated that fallacy on more than one occasion.
> >
> > No, just the facts. Try and cite a source supporting your position.
>
> Why don't you? You're the one making the claims.
>
> >
> > > Simple fact is that read performance will vary depending on server
> > > usage. RAID5 might be faster than RAID1, but it doesn't necessarily
> > > follow that it will be.
> >
> > Apples vs apples >3 drive RAID 5 will always read faster than 2 drive
> > RAID 1.
> >
> > > Bear in mind that in RAID5, the parity information must also be
> > > read,
> >
> > FALSE!
>
> Really? So how does the RAID controller know that the read data is
> correct if it doesn't read the parity information too.

How is any disk read determined to be valid data? You were the one who said
that RAID 1 allows simultaneous independent reads from both drives thus
precluding reading the data twice and comparing for vailidity checking. If
you've studied storage and disk technology then you'd know that a HD has
extensive ECC encoding on the drive's surface thus precluding invalid reads
except once every giant meteroid impact cycle. That per disk intrinsic ECC
is what RAID and any other disk depends on during all reads except when a
drive or any portion thereof has failed.

> What on earth
> would be the point of calculating and writing parity data if it's not
> going to be used?

HUH? If is obviously for the purpose of redundantly recording the data for
when ONE of the HDs fails and NOT to make general disk access SLOWER!

> > > so more data than needed must be read.
> >
> > Do you actually know anything about how RAID 5 works?
>
> Do you?

Obviously!

> > > Bear in mind that in RAID1, two read requests can be satisfied
> > > simultaneously by applying one to each half of the mirror.
> >
> > Right.
> >
> > > RAID5 cannot
> > > do that - each request must be satisfied in turn.
> >
> > Flat WRONG. RAID 5 could have an independent read going on EACH
> > drive in the array simultaneously and under load WILL.
>
> Ahem, in RAID 5, each block of data is spread across multiple drives,

'Block'....what's a block in this context? Do you know the difference
between blocked RAID 5 and striped RAID 5? Do you know the difference
between a cluster, a stripe and a stripe unit? What entity exactly do you
believe is "spread across multiple drives" ?

> so to complete any read request on RAID5, all drives must be read
> together.

That's flat false whether blocked or striped .

A striped RAID 5 read is basically the same as a RAID 0 read.

> This is how RAID5 manages to survive any one drive failing -
> because both the data and the parity information is spread across all
> drives.

That's right. Now go study up on exactly how that's done rather than just
guessing.

> > > Bottom line, a RAID5 array with just 3 or 4 drives is unlikely to
> > > outperform a RAID1 mirror under general I/O conditions.
> >
> > Bottom line is that you obviously have little experience/knowledge
> > regarding storage technology.
>
> Bottom line, you really have no clue.

Obviously you've been outclassed.

> > > However, in situations where the data requested is all contiguous
> > > and in large quantities, a RAID5 array probably would beat a RAID1
> > > mirror, even with just 3 drives.
> >
> > Well, >3 drive RAID 5 will be faster that 2 drive RAID 1 in small
> > record random I/O situations as well.
>
> No, because the read requests must be queued on RAID5, where they can
> be executed in parallel on RAID1.

That's flat false. All drives in a RAID 5 array may be doing independent
I/Os simultaneously. Both cases allow 'parallel' operation. In a
busy(saturated) SCSI RAID 5 array each drive WILL have MANY I/Os queued
onboard each drive simultaneously. They get there because the device
driver, OS and RAID controller firmware all know the array can be fed many
I/Os before the first one finishes. Study up on the concept of HD 'command
queuing' and how it's exploited in RAID 5 arrays.

> > > Given that an SBS is a general-purpose server, RAID10 probably
> > > offers the best all-round performance,
> >
> > Disreputable jump to RAID 10, I see.
>
> And why would mention of RAID10 be "disreputable"? I guess you're not
> comfortable with using it, ergo, it must be a bad choice.

Nice try but you were transparently failing in your argument of RAID 1 vs
RAID 5 so you tried to sneak in RAID 10.

> > The best all around performance in RAID 0 but that is anti-reliable.
> > RAID 5 performs like RAID 0 on reads but with one fewer drives.
>
> What is this drivel? RAID5 has the overhead of parity information,
> which is not the case with RAID1 or RAID10.

You simply have NO CLUE regarding storage and array technology. That modest
RAID 5 overhead only is suffered on WRITES!

> > RAID 10 is excessively expensive in that it requires double the
> > number of drives.
>
> That depends on your POV. But the question of cost is not directly
> relevant to a discussion of performance.

There's only price performance.

> Now if we want to start
> talking about price/performance ratios and cost considerations, then
> yes, we can start talking about RAID10 being more expensive than RAID5
> or RAID1.
>
> >
> > > RAID1 probably next best, and RAID5
> > > probably comes 3rd.
> >
> > RAID 5 is the preferred solution when one needs more than one drive
> > except where there is a specific WRITE performance requirement that
> > one is willing to pay the RAID 10 premium to get.
>
> RAID5 is one possible solution. It may be *your* preferred solution in
> all cases, but I prefer to weigh the merits of each depending on the
> situation.

Your customers would prefer that you retain someone competent in storage
technology to assist them with their selections.

> > Study up on storage technology.
>
> Study up on social skills.

And when we can't carry the day technically we always digress to personal
attacks.

Study up on storage technology.

Relevant Pages

  • Re: Problems with software RAID on SATA
    ... Connected to this are two 320GB drives ... >>which I want to turn into a RAID1 array. ... >>I'm almost certain it's a problem with initting the RAID arrays at boot. ...
    (Debian-User)
  • Re: RAID newbie...can I have several partitions on a RAID 1 array?
    ... You haven't expounded upon why you think you need raid. ... better backup device rather than buy 2 cheap RAID HBAs. ... RAID array then I would have to replace the mobo with the same one or at ... Lets say, for example, you buy 2 identical model drives, from ...
    (comp.sys.ibm.pc.hardware.storage)
  • Re: [PATCH 000 of 5] md: Introduction
    ... "why linux raid isn't Raid really, why it can be worse than plain disk") ... After this, the array ... error is in the filesystem, due to the complex layout of raid5. ... hundreds or 1000s of drives, you've quite high probability that some of them will fail sometimes, or will develop a bad sector etc). ...
    (Linux-Kernel)
  • Corrupt Win2k Software RAID 5 PLEASE HELP **
    ... I have an IDE raid 5 array implemented with windows 2000. ... The OS is on a stand alone disk. ... One of the hard drives ...
    (microsoft.public.win2000.setup)
  • Re[2]: Whay is broken ATARAID that ignored?
    ... > That's not what I call a working raid support! ... For all I know that's actually a bug in the controller firmware (there ... But rather this than FreeBSD going on to write to a RAID1 array ... drives, modify just one and then expect it to still work. ...
    (freebsd-current)