pg.ddx.io pgsql-performance@postgresql.org mailing list archive
help / color / mirror / Atom feedUltra-cheap NVRAM device
9+ messages / 8 participants
[nested] [flat]
* Ultra-cheap NVRAM device
@ 2005-10-03 11:02 Steinar H. Gunderson <sgunderson@bigfoot.com>
2005-10-03 17:15 ` Re: Ultra-cheap NVRAM device Dan Harris <fbsd@drivefaster.net>
2005-10-03 17:21 ` Re: Ultra-cheap NVRAM device Dan Harris <fbsd@drivefaster.net>
2005-10-03 19:06 ` Re: Ultra-cheap NVRAM device Vivek Khera <vivek@khera.org>
2005-10-04 21:02 ` Re: Ultra-cheap NVRAM device Jim C. Nasby <jnasby@pervasive.com>
0 siblings, 4 replies; 9+ messages in thread
From: Steinar H. Gunderson @ 2005-10-03 11:02 UTC (permalink / raw)
To: pgsql-performance
I thought this might be interesting, not the least due to the extremely low
price ($150 + the price of regular DIMMs):
http://www.tomshardware.com/storage/20050907/index.html
Anybody know a good reason why you can't put a WAL on this, and enjoy a hefty
speed boost for a fraction of the price of a traditional SSD? (Yes, it's
SATA, not PCI, so the throughput is not all that impressive -- but still,
it's got close to zero seek time.)
/* Steinar */
--
Homepage: http://www.sesse.net/
^ permalink raw reply [nested|flat] 9+ messages in thread
* Re: Ultra-cheap NVRAM device
2005-10-03 11:02 Ultra-cheap NVRAM device Steinar H. Gunderson <sgunderson@bigfoot.com>
@ 2005-10-03 17:15 ` Dan Harris <fbsd@drivefaster.net>
2005-10-03 17:36 ` Re: Ultra-cheap NVRAM device Jeffrey W. Baker <jwbaker@acm.org>
3 siblings, 1 reply; 9+ messages in thread
From: Dan Harris @ 2005-10-03 17:15 UTC (permalink / raw)
To: pgsql-performance
On Oct 3, 2005, at 5:02 AM, Steinar H. Gunderson wrote:
> I thought this might be interesting, not the least due to the
> extremely low
> price ($150 + the price of regular DIMMs):
>
>
>
This has been posted before, and the main reason nobody got very
excited is that:
a) it only uses the PCI bus to provide power to the device, not for I/O
b) It is limited to SATA bandwidth
c) The benchmarks did not prove it to be noticeably faster than a
good single SATA drive
A few of us were really excited at first too, until seeing the
benchmarks..
-Dan
^ permalink raw reply [nested|flat] 9+ messages in thread
* Re: Ultra-cheap NVRAM device
2005-10-03 11:02 Ultra-cheap NVRAM device Steinar H. Gunderson <sgunderson@bigfoot.com>
2005-10-03 17:15 ` Re: Ultra-cheap NVRAM device Dan Harris <fbsd@drivefaster.net>
@ 2005-10-03 17:36 ` Jeffrey W. Baker <jwbaker@acm.org>
0 siblings, 0 replies; 9+ messages in thread
From: Jeffrey W. Baker @ 2005-10-03 17:36 UTC (permalink / raw)
To: pgsql-performance
On Mon, 2005-10-03 at 11:15 -0600, Dan Harris wrote:
> On Oct 3, 2005, at 5:02 AM, Steinar H. Gunderson wrote:
>
> > I thought this might be interesting, not the least due to the
> > extremely low
> > price ($150 + the price of regular DIMMs):
> >
> >
> >
>
> This has been posted before, and the main reason nobody got very
> excited is that:
>
> a) it only uses the PCI bus to provide power to the device, not for I/O
> b) It is limited to SATA bandwidth
> c) The benchmarks did not prove it to be noticeably faster than a
> good single SATA drive
>
> A few of us were really excited at first too, until seeing the
> benchmarks..
Also, no ECC support. You'd be crazy to use it for anything.
-jwb
^ permalink raw reply [nested|flat] 9+ messages in thread
* Re: Ultra-cheap NVRAM device
2005-10-03 11:02 Ultra-cheap NVRAM device Steinar H. Gunderson <sgunderson@bigfoot.com>
@ 2005-10-03 17:21 ` Dan Harris <fbsd@drivefaster.net>
2005-10-05 17:01 ` Re: Ultra-cheap NVRAM device Chris Browne <cbbrowne@acm.org>
3 siblings, 1 reply; 9+ messages in thread
From: Dan Harris @ 2005-10-03 17:21 UTC (permalink / raw)
To: pgsql-performance
On Oct 3, 2005, at 5:02 AM, Steinar H. Gunderson wrote:
> I thought this might be interesting, not the least due to the
> extremely low
> price ($150 + the price of regular DIMMs):
>
>
Replying before my other post came through.. It looks like their
benchmarks are markedly improved since the last article I read on
this. There may be more interest now..
-Dan
^ permalink raw reply [nested|flat] 9+ messages in thread
* Re: Ultra-cheap NVRAM device
2005-10-03 11:02 Ultra-cheap NVRAM device Steinar H. Gunderson <sgunderson@bigfoot.com>
2005-10-03 17:21 ` Re: Ultra-cheap NVRAM device Dan Harris <fbsd@drivefaster.net>
@ 2005-10-05 17:01 ` Chris Browne <cbbrowne@acm.org>
0 siblings, 0 replies; 9+ messages in thread
From: Chris Browne @ 2005-10-05 17:01 UTC (permalink / raw)
To: pgsql-performance
fbsd@drivefaster.net (Dan Harris) writes:
> On Oct 3, 2005, at 5:02 AM, Steinar H. Gunderson wrote:
>
>> I thought this might be interesting, not the least due to the
>> extremely low
>> price ($150 + the price of regular DIMMs):
>
> Replying before my other post came through.. It looks like their
> benchmarks are markedly improved since the last article I read on
> this. There may be more interest now..
It still needs a few more generations worth of improvement.
1. It's still limited to SATA speed
2. It's not ECC smart
What I'd love to see would be something that much smarter, or, at
least, that pushes the limits of SATA speed, and which has both a
battery on board and enough CF storage to cope with outages.
--
output = reverse("gro.mca" "@" "enworbbc")
http://www.ntlug.org/~cbbrowne/linuxxian.html
We all live in a yellow subroutine, a yellow subroutine, a yellow
subroutine...
^ permalink raw reply [nested|flat] 9+ messages in thread
* Re: Ultra-cheap NVRAM device
2005-10-03 11:02 Ultra-cheap NVRAM device Steinar H. Gunderson <sgunderson@bigfoot.com>
@ 2005-10-03 19:06 ` Vivek Khera <vivek@khera.org>
3 siblings, 0 replies; 9+ messages in thread
From: Vivek Khera @ 2005-10-03 19:06 UTC (permalink / raw)
To: pgsql-performance
On Oct 3, 2005, at 7:02 AM, Steinar H. Gunderson wrote:
> Anybody know a good reason why you can't put a WAL on this, and
> enjoy a hefty
> speed boost for a fraction of the price of a traditional SSD? (Yes,
> it's
> SATA, not PCI, so the throughput is not all that impressive -- but
> still,
> it's got close to zero seek time.)
>
old news. discussed here a while back.
the board you see has no ECC. Would you trust > 1GB RAM to not have
ECC for more than 1 month? You're almost guaranteed at least 1 bit
error.
^ permalink raw reply [nested|flat] 9+ messages in thread
* Re: Ultra-cheap NVRAM device
2005-10-03 11:02 Ultra-cheap NVRAM device Steinar H. Gunderson <sgunderson@bigfoot.com>
@ 2005-10-04 21:02 ` Jim C. Nasby <jnasby@pervasive.com>
3 siblings, 0 replies; 9+ messages in thread
From: Jim C. Nasby @ 2005-10-04 21:02 UTC (permalink / raw)
To: pgsql-performance
There was a discussion about this about 2 months ago. See the archives.
On Mon, Oct 03, 2005 at 01:02:26PM +0200, Steinar H. Gunderson wrote:
> I thought this might be interesting, not the least due to the extremely low
> price ($150 + the price of regular DIMMs):
>
> http://www.tomshardware.com/storage/20050907/index.html
>
> Anybody know a good reason why you can't put a WAL on this, and enjoy a hefty
> speed boost for a fraction of the price of a traditional SSD? (Yes, it's
> SATA, not PCI, so the throughput is not all that impressive -- but still,
> it's got close to zero seek time.)
>
> /* Steinar */
> --
> Homepage: http://www.sesse.net/
>
> ---------------------------(end of broadcast)---------------------------
> TIP 3: Have you checked our extensive FAQ?
>
> http://www.postgresql.org/docs/faq
>
--
Jim C. Nasby, Sr. Engineering Consultant jnasby@pervasive.com
Pervasive Software http://pervasive.com work: 512-231-6117
vcard: http://jim.nasby.net/pervasive.vcf cell: 512-569-9461
^ permalink raw reply [nested|flat] 9+ messages in thread
* Re: Ultra-cheap NVRAM device
@ 2005-10-03 18:00 Ron Peacetree <rjpeace@earthlink.net>
0 siblings, 0 replies; 9+ messages in thread
From: Ron Peacetree @ 2005-10-03 18:00 UTC (permalink / raw)
To: Dan Harris <fbsd@drivefaster.net>; pgsql-performance
Nah. It's still not right. It needs:
1= full PCI, preferably at least 64b 133MHz PCI-X, bandwidth.
A RAM card should blow the doors off the fastest commodity
RAID setup you can build.
2= 8-16 DIMM slots
3= a standard battery type that I can pick up spares for easily
4= ECC support
If it had all those features, I'd buy it at even 2x or possibly
even 3x it's current price.
8, 16, or 32GB (using 1, 2, or 4GB DIMMs respectively in an 8 slot
form factor) of very fast temporary work memory (sorting
anyone ;-) ). Yum.
Ron
-----Original Message-----
From: Dan Harris <fbsd@drivefaster.net>
Sent: Oct 3, 2005 1:21 PM
To: pgsql-performance@postgresql.org
Subject: Re: [PERFORM] Ultra-cheap NVRAM device
On Oct 3, 2005, at 5:02 AM, Steinar H. Gunderson wrote:
> I thought this might be interesting, not the least due to the
> extremely low
> price ($150 + the price of regular DIMMs):
>
>
Replying before my other post came through.. It looks like their
benchmarks are markedly improved since the last article I read on
this. There may be more interest now..
-Dan
---------------------------(end of broadcast)---------------------------
TIP 6: explain analyze is your friend
^ permalink raw reply [nested|flat] 9+ messages in thread
* Re: Ultra-cheap NVRAM device
@ 2005-10-05 18:35 Merlin Moncure <merlin.moncure@rcsonline.com>
0 siblings, 0 replies; 9+ messages in thread
From: Merlin Moncure @ 2005-10-05 18:35 UTC (permalink / raw)
To: Chris Browne <cbbrowne@acm.org>; +Cc: pgsql-performance
Chris wrote:
> fbsd@drivefaster.net (Dan Harris) writes:
> > On Oct 3, 2005, at 5:02 AM, Steinar H. Gunderson wrote:
> >
> >> I thought this might be interesting, not the least due to the
> >> extremely low
> >> price ($150 + the price of regular DIMMs):
> >
> > Replying before my other post came through.. It looks like their
> > benchmarks are markedly improved since the last article I read on
> > this. There may be more interest now..
>
> It still needs a few more generations worth of improvement.
>
> 1. It's still limited to SATA speed
> 2. It's not ECC smart
3. Another zero (or two) on the price tag :). While it looks like a fun
toy to play with, for it to replace hard drives in server environments
they need to provide more emphasis and effort in assuring people their
drive is reliable.
If they really wanted it to be adopted in server environments, it would
have been packaged in a 3.5" drive, not a pci card, since that's what we
all hot swap (especially since it already uses SATA interface). They
would also have allowed use of 2 and 4gb DIMS, and put in a small hard
drive that the memory paged to when powered off, and completely isolated
the power supply...hard to pack all that in 60$.
That said, we are in the last days of the hard disk. I think it is only
a matter of months before we see a sub 1000$ part which have zero
latency in the 20-40 GB range. Once that happens economies of scale
will kick in and hard drives will become basically a backup device.
Merlin
^ permalink raw reply [nested|flat] 9+ messages in thread
end of thread, other threads:[~2005-10-05 18:35 UTC | newest]
Thread overview: 9+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
2005-10-03 11:02 Ultra-cheap NVRAM device Steinar H. Gunderson <sgunderson@bigfoot.com>
2005-10-03 17:15 ` Dan Harris <fbsd@drivefaster.net>
2005-10-03 17:36 ` Jeffrey W. Baker <jwbaker@acm.org>
2005-10-03 17:21 ` Dan Harris <fbsd@drivefaster.net>
2005-10-05 17:01 ` Chris Browne <cbbrowne@acm.org>
2005-10-03 19:06 ` Vivek Khera <vivek@khera.org>
2005-10-04 21:02 ` Jim C. Nasby <jnasby@pervasive.com>
2005-10-03 18:00 Re: Ultra-cheap NVRAM device Ron Peacetree <rjpeace@earthlink.net>
2005-10-05 18:35 Re: Ultra-cheap NVRAM device Merlin Moncure <merlin.moncure@rcsonline.com>
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox