OCZ Z-Drive 1TB SSD Performance Preview at Cebit

http://www.madshrimps.be/?action=g [...] rticID=911


Did anybody else notice the "apples-and-oranges" comparisons here?

Notably, they measured:

4 x 256GB OCZ SSD in RAID 0
with Highpoint RocketRAID 3520
(presumably SATA @ 300MB/sec)

4 x 36GB Raptor in RAID 0
(SATA fixed at 150MB/sec;
controller not mentioned)

1 x Gigabyte i-RAM, no RAID
(SATA fixed at 150MB/sec;
controller not mentioned)

2 x Maxtor 73GB 15,000 rpm in RAID 0
(SCSI-4 interface;
controller not mentioned)

2 x 36GB Raptor in RAID 0
(SATA fixed at 150MB/sec;
controller not mentioned)


Conclusion: I can build the
Z-drive myself with that
Highpoint RocketRAID 3520
and SSDs of my own choosing.
But, there is little point in
comparing such a setup with
1 or 2-drive arrays and
without mentioning the most
crucial component: the controller!!

WHAT'S WRONG WITH THESE REVIEWERS?

MRFS

> if they are gunna compare a pcie product how about to a similar product.
> http://www.fusionio.com/Products.aspx


RIIIIIGHT!

http://www.fusionio.com/PDFs/Fusion%20Specsheet.pdf

... says that all 3 capacities use an x4 PCI-Express slot.


So, a better review would have compared x4 and x8 slots,
with corresponding lane assignments by the BIOS and chipset,
in order to help isolate the limiting factor: bus, controller or memory.

Also, some chipsets only assign x4 lanes to x8 mechanical slots,
so "caveat emptor" (buyer beware).


This next editorial is also interesting, because it WRONGLY assumed
that a single RAID controller was bound by x8 PCI-Express lanes
@ 250MB/second = 2,000 MB/second:

http://www.tomshardware.com/news/s [...] ,7224.html

However, as the comments pointed out, those SSDs were
spread across 3 different RAID controllers:

they had
10 drives connected to an 'areca 1680ix-24'
8 to an 'adaptec 5 series'
and 4 directly to the mobo.

[end quote]


Nevertheless, I still maintain that storage subsystems
should NOW be getting the same bandwidth attention
that SLI and CROSSFIRE video setups have been
getting since the dawn of PCI-Express.

For example, Highpoint makes RAID controllers
that support "teaming" -- 2 x PCI-E controllers
installed in separate PCI-E slots.

So, why are RAID controller manufacturers
not working on models that fully exploit
either all x16 PCI-E 1.0 lanes, or x8 PCI-E 2.0 lanes?

Maybe all the attention SSDs are getting will
motivate them to realize the potential
of such a high-end RAID controller.

When I realized the OCZ Z-drive is assembled with
components I can buy separately, a light went on
in my head


Highpoint, are you reading this, by any chance?
Areca? 3Ware? Adaptec? Anybody??


MRFS

No. I was complaining about the "apples-and-oranges" comparisons, chiefly e.g.:

4 x 256GB OCZ SSD in RAID 0
with Highpoint RocketRAID 3520
(presumably SATA @ 300MB/sec)

vs.

1 x Gigabyte i-RAM, no RAID
(SATA fixed at 150MB/sec;
controller not mentioned)


All the other devices were rotating disk drives.


MRFS