[ale] Terabyte of data

hbbs at attbi.com hbbs at attbi.com
Wed Feb 26 08:57:39 EST 2003


I remembered this from Slashdot:

http://www.kcgeek.com/content/features/1011742784.Peloquin.Hardware/feature.html

This design dispenses with any kind of hardware RAID, favoring cheap PCI ATA 
controllers and performing RAID in software.

This cries out for dual fast CPUs since they'll be generating parity under a 
RAID 5 regime; you can always overclock and watercool if it comes down to 
that.  You will probably want to find a specialty case with dual power supplies 
and lots of 5-1/4" drive bays, front-accessible.

When I first saw this article, I think it was longer than this, but as I 
recall, they went ahead and stuck on two drives per bus and damn the speed 
hit.  You may want to try this out and/or think it over.  This may not be a big 
deal if your particular application can make good use of disk cache and you 
throw in at least 1GB of RAM.  

Make sure you look at www.raidzone.com.  What I don't seem to be seeing there 
anymore are controller cards (closed driver for Linux IIRC) and external 
arrays; you'd have to call and see (they've been very good over the phone in 
the past).

- Jeff
> My main concern is processor speed.  I'm going to be providing this data
> as well as writing software to keep it Blowfish encrypted.  Every bit of
> it.  I do not know the specs of why/how the blowfish has to work but I
> know that it may need to be done.  I need good product because I need a
> "cookie cutter" solution.  My main concern is FS speed and hardware
> speed.  I'll be investigating which FS can give use good speed.  It may
> be that I would have to write a very small FS that provides the basics
> and does encryption on the fly.  I need at least 128bit encryption.  The
> more I think about it, it may be a custom fs is needed.
> 
> 
> 
> On Wed, 2003-02-26 at 07:28, Robert L. Harris wrote
> > 
> > 
> > I'm currently using a system that has 16 160G disks in a RAID5.  This 
> > gives me a bit more than you want of course.  I'm using a 3Ware IDE Raid 
> > card to provide connectivity but using Software RAID5.  Back when we
> > built this system the 3ware RAID5 implementation was a disaster for
> > performance, this may have changed however.
> > 
> > One note, with the 3ware card, the IDE drives are hot swapable if you
> > get a system with the right type of bays.
> > 
> > I can give you the name of our vendor though personally I'd recomend a
> > local vendor as ours is in canada and I'm having issues with him.   If
> > you go with someone local who has questions on design let me know, I'll
> > give exact specs on the system/chasis we're using.  The only thing I
> > recomend is a better system for locking down the drives as sometimes the
> > drives in our system will unseat themselves.....  Also, more fans, these
> > produce a good bit of heat.  Otherwise we have 8 of these providing ALOT
> > of disk space for our legato backup server and they're working very well
> > after a good bit of tuning.
> > 
> > Robert
> > 
> > 
> > 
> > 
> > Thus spake cfowler (cfowler at outpostsentinel.com):
> > 
> > > 
> > > I'm looking at a situation where I need to create a raid 0 file system that
> > > equals 1 terabyte of storeage for trival data.  Has anyone done this with
> > > Linux?  I think the ext2 fs does have a limit that may keep it from going up
> > > to a terabyte in size.  I'll also have a limit on the file sizes too. 
> > > 
> > > Is there an ATA raid card that can do more than 4 drives.  It seems the
> > > biggest I can get is 200Mb drives.  I'll need at least 5 maybe 6 drives.
> > > _______________________________________________
> > > Ale mailing list
> > > Ale at ale.org
> > > http://www.ale.org/mailman/listinfo/ale
> > 
> > :wq!
> > ---------------------------------------------------------------------------
> > Robert L. Harris                     | PGP Key ID: E344DA3B
> >                                          @ x-hkp://pgp.mit.edu 
> > DISCLAIMER:
> >       These are MY OPINIONS ALONE.  I speak for no-one else.
> > 
> > Diagnosis: witzelsucht
> -- 
> "The Law of Leaky Abstractions"
> There is a time where abstractions lead to the inablity to 
> fix problems that leak through the abstraction.
> http://www.joelonsoftware.com/articles/LeakyAbstractions.html
> 
> _______________________________________________
> Ale mailing list
> Ale at ale.org
> http://www.ale.org/mailman/listinfo/ale
_______________________________________________
Ale mailing list
Ale at ale.org
http://www.ale.org/mailman/listinfo/ale






More information about the Ale mailing list