Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!rutgers!att!mtuxo!mtgzz!drutx!druhi!warn
From: warn@druhi.ATT.COM (WarnerD)
Newsgroups: comp.realtime
Subject: Re: fast storage?
Message-ID: <4239@druhi.ATT.COM>
Date: 15 May 89 18:25:59 GMT
References: <DOVE.89May10131213@rocket.UUCP> <1264@mmm.UUCP> <5427@videovax.tv.Tek.com> <3615@eos.UUCP>
Reply-To: warn@druhi.ATT.COM (WarnerD)
Distribution: comp
Organization: AT&T, Denver, CO
Lines: 64

In article <3615@eos.UUCP> eugene@eos.UUCP (Eugene Miya) writes:
>In article <5427@videovax.tv.Tek.com> bart@videovax.tv.tek.com (Bart Massey) writes:
>>In article <1264@mmm.UUCP> schultz@mmm.UUCP (John C Schultz) writes:
>>*> In article <DOVE.89May10131213@rocket.UUCP> dove@rocket.UUCP (Webster Dove) writes:
>>*> >Is there any faster large capacity storage that a Fujitsu 2380A
>>*> >parallel head 1Gbyte disk (15 Mbyte/sec) that could be used to store
>>*> >strictly sequential data?
>>*> 
>>*> A moderately priced disk drive maybe.  Pretty high speed for tape.  
>>
>>This sounds like a made-to-order application for disk-striping.  Should you
>>be considering 10 100M drives?  The most recent Usenix proceedings have
>>some interesting articles on disk farms for UNIX systems...
>
>It's all tradeoffs by cost, (custom or otherwise), and requirements.
>There is the Cray SSD with GB/second transfers (and you get random access)
>and GW capacity, but it's no micro.  And there is the TMC Data Vault.
>If you have to ask the price, you can't afford.
>
>Longish signature follows "Type 'n' now"
>
>Another gross generalization from
>
>--eugene miya, NASA Ames Research Center, eugene@aurora.arc.nasa.gov
>  resident cynic at the Rock of Ages Home for Retired Hackers:
>  "You trust the `reply' command with all those different mailers out there?"
>  "If my mail does not reach you, please accept my apology."
>  {ncar,decwrl,hplabs,uunet}!ames!eugene
>  				Live free or die.

In a recent past life I was manager of software development for Seagate
Technology's now deceased but possibly reincarnated (as Array Technology Corp.) disk array project located in Boulder, CO.  We developed a prototype that was
privately shown in a suite at Fall '88 Comdex.  As a result, I'm partial
to array architectures as a solution to high capacity, high reliability,
high bandwidth, low cost storage.  There are advantages inherent in
arrays over single spindle PTD's such as Fuji or Ibis drives.

This isn't a commercial (since the product is not now, and quite likely
never will be sold - too bad!) but a brief description of the Seagate
array's capabilities might be in order as a basis for comparison, etc.

	- up to 30GB in 3 19" racks; could be shrunk, "only" packaging

	- up to 30MB/sec net transfer rate to/from host - not many hosts
	  can handle that but the array WON'T be the bottleneck

	- some awfully large MTBDL numbers (>1,000,000 hrs) given usual
	  individual drive failure rates, the usual replacement assumptions,
	  etc.

	- a cold sparing feature which, when combined with the two drive
	  error correction, would allow three drives to fail before any
	  data was inaccessible

	- a very flexible set of configuration options, i.e. virtual volume
	  capacities, high transaction or high bandwidth mode, maintenance
	  activities, usage statistics, etc.

	- approx. $7-$9/MB net "plug it into your system" cost - VME i/f
	  was built but HyperChannel, HSC, ,IPI-3, FDDI things, etc were being
	  investigated.

I could go on for $'s and $'s on this topic, but ...  I'd enjoy corresponding 
on this subject if anyone would care to.