- Fastest workstation storage
- Posted by .::SuperBLUE::. on April 24th, 2005
Hello,
I am looking for fastest raid 0 storage solution for a workstation. Is
Adaptec 39320 controler + 6 Fujitsu mas 36gb 15000 rpm drives u 320 the
fastest i could get or is there something else, faster? Thanks
- Posted by Peter on April 24th, 2005
Fast is a generic term. Be more specific. Fast RAID for a
database, web server or a streaming application requires a
different selection criteria.
- Posted by calypso@fly.srk.fer.hr.invalid on April 24th, 2005
U comp.periphs.scsi .::SuperBLUE::. <superblue@dontusefakkevip.hr> prica:
39320 is not a RAID controller... You could possibly use software RAID0
solution, but I don't recommend it... Here is a list of Adaptec's RAID
controllers:
http://www.adaptec.com/worldwide/sup...e=supportindex
Also check LSI and ICP Vortex...
Second... What kind of speed you need? Fast read, write, seek?
--
Tapira spava sretan kaktusu maltretiru danima u saboru.
By runf
Damir Lukic, calypso@_MAKNIOVO_fly.srk.fer.hr
a member of hr.comp.hardver FAQ-team
- Posted by Eric Gisin on April 24th, 2005
Don't be stupid. Hardware RAID is of no benefit on the desktop, especially
from Adaptec.
If you do not have PCI-X, the fastest (STR) available is RAID 0 on Intel's
i9XX and nForce4 SATA2.
If you want a fast workstation, get a 15K SCSI OS drive, and IDE for data.
<calypso@fly.srk.fer.hr.invalid> wrote in message
news:d4g3q8$4pe$3@bagan.srce.hr...
nglish+US&cat=%2FTechnology%2FRAID&fromPage=suppor tindex
- Posted by Folkert Rienstra on April 24th, 2005
<calypso@fly.srk.fer.hr.invalid> wrote in message news:d4g3q8$4pe$3@bagan.srce.hr
Yes, it is. Just not an all purpose one.
That is what all RAID is, software.
The only difference is where this software runs on/runs in,
whether that's a dedicated processor, bios+driver or driver only.
Who cares what you recommend.
Do you always have to ask questions that the post already answers?
- Posted by Folkert Rienstra on April 24th, 2005
"Eric Gisin" <ericgisin@hotmail.com> wrote in message news:d4gcdv01sur@enews2.newsguy.com
And you said "Don't be stupid"?
- Posted by .::SuperBLUE::. on April 24th, 2005
You could possibly use software RAID0
Why not? Will it make a serious impact on performance of a dual opteron.
Fastest possible - read, write, seek.
http://www.fcpa.fujitsu.com/products...specifications.
html
- Posted by .::SuperBLUE::. on April 24th, 2005
This workstation is used for various 35mm film and hdtv video production
effects.
- Posted by .::SuperBLUE::. on April 24th, 2005
- Posted by J. Clarke on April 24th, 2005
..::SuperBLUE::. wrote:
If your only criterion is "the fastest possible" then you want a RAM based
device, not disk.
--
--John
to email, dial "usenet" and validate
(was jclarke at eye bee em dot net)
- Posted by J. Clarke on April 24th, 2005
..::SuperBLUE::. wrote:
Depends on your definition. It appears to be host-based, in other words
it's software RAID only with BIOS support for boot.
--
--John
to email, dial "usenet" and validate
(was jclarke at eye bee em dot net)
- Posted by Folkert Rienstra on April 24th, 2005
"Peter" <peterfoxghost@yahoo.ca> wrote in message news:l0Mae.71$BW6.9170@news20.bellglobal.com
Oh? How so?
- Posted by Neill Massello on April 25th, 2005
..::SuperBLUE::. <superblue@dontusefakkevip.hr> wrote:
So you need fast reads and writes on large files, but not extremely fast
access. For this application, the high (15k) rpm drives you mentioned
won't provide much additional benefit to offset their high cost.
- Posted by J. Clarke on April 25th, 2005
Maxim S. Shatskih wrote:
Unless your main process is CPU-bound.
--
--John
to email, dial "usenet" and validate
(was jclarke at eye bee em dot net)
- Posted by Maxim S. Shatskih on April 25th, 2005
Software RAIDs are often more reliable and better coded then hardware ones. The
Linux community or Veritas Corp (MS licensed the lite version of Veritas's VxVM
to Windows as "Dynamic Disk") can produce better code then the small company
with small development bugdets and heavy time-to-market requirements.
The only drawbacks of the software RAID are a) hardships in some OSes like
Windows in booting from it - Windows can boot from mirror only, not from stripe
or span set or RAID5 b) RAID5 checksum calculation is done by the host CPU,
which is slow.
If one does not use RAID5, but uses mirror - then I see no advantages in
hardware RAID.
--
Maxim Shatskih, Windows DDK MVP
StorageCraft Corporation
maxim@storagecraft.com
http://www.storagecraft.com
- Posted by Nick Payne on April 25th, 2005
My experiments with Win2003 server backed by EMC CX-700 (2Gbit
fibre-channel) show that best throughput was achieved with hardware RAID-10
sets striped in software so that multiple LUNs were visible to the OS. With
a 2Tb drive created from five 400Gb RAID-10 sets each consisting of six
133Gb drives, throughput was several hundred percent better than with the
RAID-10 all done in hardware and only presenting a single LUN to the OS.
OTOH, with Win2000 as the OS the software striping gave no appreciable
advantage over the hardware striping.
My testing was done with the MS Loadsim 2003 stress test tool for simulating
a production Exchange environment.
Nick
- Posted by Eric Gisin on April 25th, 2005
"Folkert Rienstra" <see_reply-to@myweb.nl> wrote in message
news:426c3082$0$82665$892e7fe2@authen.white.readfr eenews.net...
application, because of price and drive capacity.
4 IDE drives will do 300MB/s today, an Ultra320 channel is less.
Tell us Folknuts, what would you recommend? Or are you just going to troll?
- Posted by Bill Todd on April 25th, 2005
Maxim S. Shatskih wrote:
....
Well, if your controller has even a very small amount of NVRAM in which
to record current write operations, the hardware approach can avoid the
need to perform a complete mirror resynch after something as common as a
power failure (even when such software resynchs are done in the
background, they consume disk performance resources). And if your
controller has somewhat more (preferably mirrored) NVRAM, it can perform
some significant write-back optimizations on the write stream (while
significantly improving perceived write latency in the process).
Other than that, and avoiding the additional processor interrupts and
bus and memory activity which software mirroring requires, I agree.
- bill
- Posted by calypso@fly.srk.fer.hr.invalid on April 25th, 2005
U comp.periphs.scsi Eric Gisin <ericgisin@hotmail.com> prica:
Well, to tell you the truth... RAID0 over 12 Hitachi 160GB/8MB 7K250 drives
(SATA), on 3Ware 9500S-12 SATA RAID controller, and I could get 350MB/s read
and 210MB/s write speed... How the heck could you get 300MB/s on 4 ATA
drives?!!!!!
3Ware 9500S-12, 128MB cache, PCI-X 66MHz
12x Hitachi 7K250, 160GB/8MB, 64kb stripe size
The machine was a video-capture workstation... Tested with BlackMagic Design
software for video-capture speed testing... Capture card was BlackMagic
Design DectLink Extreme IIRC...
--
Opekaa gura u kantini ekonomije divovski dorucaku ugnjetavu
danima. By runf
Damir Lukic, calypso@_MAKNIOVO_fly.srk.fer.hr
a member of hr.comp.hardver FAQ-team
- Posted by calypso@fly.srk.fer.hr.invalid on April 25th, 2005
U comp.periphs.scsi Folkert Rienstra <see_reply-to@myweb.nl> prica:
? Hmmm? 39320 is, accoring to Adaptec's support page plain SCSI HBA... RAID?
Sorry, I couldn't find any RAID funcion in it... Any links?
OK, I get your point...
Well, yes, who cares?
I intend to get as much info as possible...
--
"Izbacens li beogradjaninu vjesu ?" upita eurokremo tuce pajseru ubiju.
"Nisam ja nikog bombardiro !" rece kondoma pozdravlja "Ja samo rukometasu guru blesavm !"
By runf
Damir Lukic, calypso@_MAKNIOVO_fly.srk.fer.hr
a member of hr.comp.hardver FAQ-team