Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

MacRumors

macrumors bot
Original poster
Apr 12, 2001
68,122
38,886
Also updated today alongside the iBook and PowerMac releases was Apple's Xserve RAID Storage System.

The updated 3U system now offeers up to 5.6 terabytes of storage capacity at a price for storage of just over $2/GB.

The fully occupied Xserve RAID with 5.6TB of storage comes with fourteen 400GB ATA/100 drives and has a retail price of $12,999 (Apple Store).
 
Yeah...but when will they come out with the 5.6 TB iPod? :D

Seriously though, this is a good step in the right direction to get more commercial mac users on board.

-Joe
 
What is so great about 5.6TB of storage? Who has that much porn?

Seriously, the Xserve RAID is looking to be even better, but how will you format more than a few of these into one drive? The maximum capacity of Panther to format something was, as I recall, 32TB...

Still beats out WinXP...2TB! :eek:
 
Mechcozmo said:
What is so great about 5.6TB of storage? Who has that much porn?

you just need a bigger connection dude. my 3.5tb's are almost full so this'll be nice for my next one...:p;):eek:
Mechcozmo said:
Still beats out WinXP...2TB! :eek:
so longhorn req's are 2TB but winxp cant format that big...i guess that means you'll need to format after "upgrading" somehow?
 
In case anyone's wondering, the update is from 250GB drives to 400GB drives. Times fourteen :eek:

Just when MOSR was SO close to having ALL RUMORS on the front page... now it's all news! That's what comes of those pesky rumors coming true.
 
Another detail on the update - it's now Cisco and SUSE Linux certified.

It's good to see Apple playing nice in the enterprise server arena and starting to make some inroads.
 
homerjward said:
you just need a bigger connection dude. my 3.5tb's are almost full so this'll be nice for my next one...:p;):eek:

Note to self... :)

Is there a chance of Apple re-revamping HFS? We went from that to HFS+, and then to Journaled HFS+, but that is still a rather old filesystem. Will it be changed anytime soon to take advantage of the new Xserve capacities?
 
Macmaniac said:
Its a shame their not SATA drives, they should upgrade them to SATA, that would improve performance.
Uh, well now, that's not true. Each and every drive is on a separate Ultra ATA channel currently. Those Ultra ATA 7200rpm drives are going just as fast as they possilbly can (and are generally superiour drive mechanisms to ordinary off-the-shelf modules). They each include 8MB cache chips which can be activated for speed or deactivated for security. They are hot swappable to boot.

The point being that, in this context, to say they are not SATA is irrelevent.
 
Mechcozmo said:
Note to self... :)

Is there a chance of Apple re-revamping HFS? We went from that to HFS+, and then to Journaled HFS+, but that is still a rather old filesystem. Will it be changed anytime soon to take advantage of the new Xserve capacities?

Well, you could always format the filesystem with UFS. Of course, UFS has its own problems: a lot of apps won't work properly if they're installed on a UFS drive, and I don't know what the filesystem limits are in terms of sizes, etc.

But then, there are cases -- eg, where I'm working -- where you'll be hooking an Xserve RAID up to a non-Apple computer. In that situation, Apple's filesystem limits aren't an issue (but the other OS's limits might be).
 
Also new is 512+512MB controller cache standard, previously it was 128+128, with 512+512 available as a BTO upgrade.

Anyone know if there are any hardware differences other than this cache option being standard and the new drives? What if I put some 400GB drives in a previous-generation unit?
 
Good update to best-bang-for-buck product in the Apple lineup.

As for all SATA, Enterprise-grade are always behind the curve on hardware. Solid and faithful rather than blazing and risky. If it's new and risky you want two in case one fails, cause you just don't know how good they are. Next year you might see a SATA hookup, and anyway the drives themselves are the bottleneck right now. Most PATA channels don't get flooded even with a single drive per channel.
 
wrong, by a factor of 8

Mechcozmo said:
Still beats out WinXP...2TB! :eek:

http://www.microsoft.com/resources/...docs/en-us/ntfs_compared_to_fat_and_fat32.asp

Maximum volume and partition sizes start at 2 terabytes (TB) and range upward [to 256 TB]. For example, a dynamic disk formatted with a standard allocation unit size (4 KB) can have partitions of 16 TB minus 4 KB.

This is server, but realistically why would anyone put multi-TB filesystems on a desktop?

Check your "facts"... unless you're deliberately trying to mislead....
 
I'm sorry, I was not trying to mislead... apparently I was wrong. Thank you for the information.
 
I am a pretty good computer user but that kind of power just blows me away. When you get into the professional server side of things I feel like I don't even know what a computer is.
 
It's very nice, and it's a good price, but 7200RPM ATA100 drives are old tech. They should be using SATA 10k drives, or atleast SATA 7200RPM for the cheap skates :p
 
edesignuk said:
It's very nice, and it's a good price, but 7200RPM ATA100 drives are old tech. They should be using SATA 10k drives, or atleast SATA 7200RPM for the cheap skates :p
When you have fourteen hard drives, each hanging off their own controller, any interface technology that you're likely to use these days is going to be damn fast. It doesn't matter if it's ATA100, fast SCSI, SATA, or something else; the bottleneck is more than likely going to be the hard drives, not the interface.

Where I work is looking at buying four Xserve RAIDs to hook up to a system for a backup project (basically, storing gobs of data at a third site for redundancy; for our needs, tape is overkill.) As one of the guys involved in speccing this thing out, I'd be quite happy to hang the four drive boxen off a single fibre channel switch, and funnel the switch through a single fibre to the host: we're more likely to be hit by network latency and throughput issues than drive throughput and latency. (Well, ok, two fibres and two switches for redundancy, but whatever.)

The key thing here is that ATA100 is a known quantity. Serial ATA is still very new. When you're talking enterprise grade hardware, you want reliability, and you're prepared to pay for it. Reliability means that you know what the likely issues are, and what the fixes for those issues are. Give it two or three years, and yes, I'd agree that SATA is likely to make it into the Xserve RAID or an equivalent box. Not yet, though.
 
edesignuk said:
They should be using SATA 10k drives

for some reason 10krpm is bad speed for hard drives. the 10k drives tend to run hotter than they should, keep more noise than they shoud, and break easier than they shoud, and nobody knows why. as strange as it seems, 15k drives are more reliable than the slower 10k ones...
 
JFreak said:
for some reason 10krpm is bad speed for hard drives. the 10k drives tend to run hotter than they should, keep more noise than they shoud, and break easier than they shoud, and nobody knows why. as strange as it seems, 15k drives are more reliable than the slower 10k ones...
I was just using a for instance, and I said 10k becasue they would be cheaper ;)
 
This is my understanding of the performance aspects of Ultra ATA vs SATA for this setup.

The whole Xserve RAID is connected to the server/workstation with Dual 2Gb Fibre Channel ports (SFP), 200MB/s throughput per channel with guaranteed delivery (400MB/s full duplex).

Each Ultra ATA controller has a 100MB/s capacity. I don't know if that is half or full duplex. The ATA controllers' maximum throughput for a 14 disk RAID 0+5 setup with 2 parity disks will be 1200MB/s.

Currently, 7200RPM disks, like the ones in the drive modules, are capable of a sustained transfer rate of more than 30MB/s but not more than 70MB/s, so in the same RAID 0+5 setup the disks are capable of reading at a maximum rate of 840MB/s, minimum 360MB/s. Write rate will be a bit slower than read rates. Burst rates could exceed 100MB/s but not for very long.

My conclusion, the system's main bottleneck is the Fibre Channel link and in rare occasions the disks themselves, the Ultra ATA layer will almost never be the bottleneck.

The only time the ATA layer would have a challenge would be when the disks are able to deliver data from the caches on the disks, but the two RAID controllers have a cache of 512MB per unit, so the disks will never be able to provide cached data that the controllers don't already have.

Hence no need for S-ATA
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.