Re: RAID 0 risky ?

2002-03-20 Thread Jason Lim

> Some RAID cards are substantially slower at RAID 5 than 0/1/0+1.  The
> 3ware ATA RAID boards are excellent, for example, except at RAID 5.
> They now produce a couple of boards with more CPU power or different
> ASICs, or whatever, to make up for this shortfall.  But they are more
> expensive.
>
> IIRC the 7x10 series is quite slow at RAID 5, but the 7x50 series
> improved upon this greatly.  My source is www.storagereview.com, though,
> I do not use any of their newer RAID 5 boards.
>

I hear the latest 3ware 7.4 firmware upgrade has increased the performance
of RAID5 quite a bit by optimizing the read/write alogrithm. Not that I've
upgraded any of our servers... we're still on 7.3.2 (or whatever the last
7.3.x was). We'll wait till maybe 7.4.1 or until no one complaints of bugs
anymore.


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




Re: RAID 0 risky ?

2002-03-20 Thread Jeff S Wheeler

On Wed, 2002-03-20 at 01:03, Dave Watkins wrote:
> replicate the data somehow. RAID 5 obviously does the least replication 
> while still keeping fault tolerance, although it does cost a small amount 
> of computing power (not a problem if you have a RAID card)

Some RAID cards are substantially slower at RAID 5 than 0/1/0+1.  The
3ware ATA RAID boards are excellent, for example, except at RAID 5. 
They now produce a couple of boards with more CPU power or different
ASICs, or whatever, to make up for this shortfall.  But they are more
expensive.

IIRC the 7x10 series is quite slow at RAID 5, but the 7x50 series
improved upon this greatly.  My source is www.storagereview.com, though,
I do not use any of their newer RAID 5 boards.

-- 
Jeff S Wheeler   [EMAIL PROTECTED]
Software DevelopmentFive Elements, Inc
http://www.five-elements.com/~jsw/



-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




Re: RAID 0 risky ?

2002-03-20 Thread Sebastian Nerz

Hy,

Rich Puhek schrieb:
> 
> Thedore Knab wrote:
> >
 

> I'm not sure that there's significantly less work for RAID 0 than RAID
> 5. RAID 1 will definately have less reads on any particular drive than
> the other two (approzimately 50% will go to one disk, 50% to the other),
> but will have greater writes (100% of writes will affect both disks on
> RAID 1, it's possible some writes will not affect every disk on RAID 5,
> I think).
> 

As far as I know every drive will have to write some data.

An example for RAID 5:

5 HDDs (everyony 30 GB)
you will have a "resulting" disk with 120 GB of usable data ((5-1) * 30)
30 GB of Data for Parity-Checking will be divided on every HDD.
(There is no complete Parity-Disk in RAID 5 - that's the difference to
RAID 4 which *is* using a special Parity-Disk)
If you write data, this data is divided on all hard-drives.
It is more secure than RAID 0 and less secure than RAID 1.

> Since you're looking at IMAP mail files, the data is probably
> critical... too critical to trust in a single drive failure. You're
> probably most concerned with read performance (since users will notice
> lag in reading email... deleting messages can plod away just fine,
> writing to the mail files is done by the MDA). Take a look at how much
> space you need, or are likely to need in the future. If you need a lot
> of space I'd go with RAID 5. If you want to really push the read
> performance, buy another drive and go RAID 1.
> 
> I'd stay away from RAID 0 unless it's fairly non-critical data, and if
> you really need the throughput. An example I can think of would be
> something like a web cache.
> 

Yeah, but if you need high-speed you could use RAID 0+1..

(First mirror your drives, than sripe them.
This is a bit more secure :-) )



> 
> --Rich
> 
> _
> 
> Rich Puhek
> ETN Systems Inc.
> _
> 

regards

Sebastian Nerz

PS I hope the follwing links are okay:
http://whatis.techtarget.com/definition/0,289893,sid9_gci214332,00.html
http://www.win2000mag.com/Articles/Index.cfm?ArticleID=218


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




Re: RAID 0 risky ?

2002-03-19 Thread Rich Puhek


Thedore Knab wrote:
> 
> Is RAID 0 that risky anymore for data storage (IMAP mail files) ?
> 
> I figure that under normal wear and tear a drive should last about 5 years.
> 
> Does this sound right ?
> 
> I have 3 IBM SCSI 18GB drives.
> 
> With RAID 0, I get 51.5GB of storage space.
> With RAID 5, I only get 37 GB of space with 20% wasted overhead.
> 
> RAID 0 and RAID 1 are less work for the disk volume than RAID 5.
> 
> So in an ideal world, volumes with RAID 0 or RAID 1 will last longer than
> volumes in RAID 5.

No. In RAID 5, you can lose one drive, and still keep running. In RAID 0
(striped), the loss of one drive will kill your volume. Let's say the
odds are 1 in 10 of a drive dying in a particular year. If you have 5
drives in RAID 0, you have a 50% chance of your volume crashing (well,
that's probably not exactly right, but my statistics abilities are a
little rusty). If you have 5 drives in RAID 5, you have a 50% chance of
_one_ drive dying.

I'm not sure that there's significantly less work for RAID 0 than RAID
5. RAID 1 will definately have less reads on any particular drive than
the other two (approzimately 50% will go to one disk, 50% to the other),
but will have greater writes (100% of writes will affect both disks on
RAID 1, it's possible some writes will not affect every disk on RAID 5,
I think).

Since you're looking at IMAP mail files, the data is probably
critical... too critical to trust in a single drive failure. You're
probably most concerned with read performance (since users will notice
lag in reading email... deleting messages can plod away just fine,
writing to the mail files is done by the MDA). Take a look at how much
space you need, or are likely to need in the future. If you need a lot
of space I'd go with RAID 5. If you want to really push the read
performance, buy another drive and go RAID 1.

I'd stay away from RAID 0 unless it's fairly non-critical data, and if
you really need the throughput. An example I can think of would be
something like a web cache.

> 
> Thus, it would be less risk to use RAID 0 or better RAID 1 than RAID 5.

Absolutely not. RAID 0 is the highest risk. RAID 0 is actually higher
risk than a single drive.

--Rich




_
 
Rich Puhek   
ETN Systems Inc. 
_


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




Re: RAID 0 risky ?

2002-03-19 Thread Dave Watkins

Technically speaking drives don't _wear_ out... Bad sectors are generated 
because at some time the disk surface has been damaged, usually by the 
heads hitting the disk. And many faults to do with the components on the 
controller board can be traced to a poor supply of power (eg spikes and 
brownouts, a UPS will help resolve this)

Other than that, RAID 0 is more risky than a single drive as you have no 
fault tolerance (one drive fails and you lose all the data on al the 
drives), and you have three times the chance that one of the drives will 
bomb for whatever reason. Since it is for mail storage (an inherantly 
difficult data source to back up), I would say using RAID 0 would be a VERY 
bad idea, especially since you mention IMAP (eg mail stored on the server). 
If one drive fails every user you have loses their mail.

I would think RAID 5 would be the better system to use in this instance.

To follow you usage question a little. Lets assume you want to write 256k 
to the array.

(We assume 64k block size for all arrays)

In a RAID 0 situation the first drive would have 128k written and the other 
2 would have 64k written to them.

In RAID 1 (using 2 drives) each drive would have 256k written to them.

In RAID 5, each drive would have 128k written. There would be 2 x 64k 
written to the 2 data drives, as well as another 128 on the parity drive 
(for this particular write).

This is simplified but correct, from here we can see that RAID 1 would have 
the highest usage patterns per drive, next would be RAID5 and finally RAID 
0. This is of course the price you pay for redundancy, you have to 
replicate the data somehow. RAID 5 obviously does the least replication 
while still keeping fault tolerance, although it does cost a small amount 
of computing power (not a problem if you have a RAID card)

Hope this helps
Dave

At 00:09 20/03/2002 -0500, Thedore Knab wrote:
>Is RAID 0 that risky anymore for data storage (IMAP mail files) ?
>
>I figure that under normal wear and tear a drive should last about 5 years.
>
>Does this sound right ?
>
>I have 3 IBM SCSI 18GB drives.
>
>With RAID 0, I get 51.5GB of storage space.
>With RAID 5, I only get 37 GB of space with 20% wasted overhead.
>
>RAID 0 and RAID 1 are less work for the disk volume than RAID 5.
>
>So in an ideal world, volumes with RAID 0 or RAID 1 will last longer than
>volumes in RAID 5.
>
>Thus, it would be less risk to use RAID 0 or better RAID 1 than RAID 5.
>
>-
>Ted Knab
>
>
>--
>To UNSUBSCRIBE, email to [EMAIL PROTECTED]
>with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




RAID 0 risky ?

2002-03-19 Thread Thedore Knab

Is RAID 0 that risky anymore for data storage (IMAP mail files) ?

I figure that under normal wear and tear a drive should last about 5 years.

Does this sound right ?

I have 3 IBM SCSI 18GB drives. 

With RAID 0, I get 51.5GB of storage space. 
With RAID 5, I only get 37 GB of space with 20% wasted overhead. 

RAID 0 and RAID 1 are less work for the disk volume than RAID 5.

So in an ideal world, volumes with RAID 0 or RAID 1 will last longer than 
volumes in RAID 5.

Thus, it would be less risk to use RAID 0 or better RAID 1 than RAID 5.

-
Ted Knab


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]