Slashdot Mirror


University Chooses Apple RAID for Linux Cluster

An anonymous reader writes "A Linux World article describes how Swinburne University chose Apple's Xserve RAID to add storage to it's Dell linux cluster, as it was the cheapest solution. Apple was sceptical about its RAID system working with Linux, but the system was up and running in 15 minutes."

68 comments

  1. Compression - WOW! wait, I mean HOW? by seigniory · · Score: 3, Interesting

    "According to Bailes, the centre records 13TB of data per day, which gets processed in real time down to 30GB and then compressed further to 3KB."

    OK, can someone to explain to me that either yes, there is a lot of redundant data that can have crazy-good compression rates, or that no, this quote is wrong...

    1. Re:Compression - WOW! wait, I mean HOW? by MindStalker · · Score: 4, Informative

      I'm sure they meant 3GB as I know most non-normalized data generally can be compressed 10:1.

    2. Re:Compression - WOW! wait, I mean HOW? by daeley · · Score: 2, Funny

      Hey, you can store a lot of compressed pictures of polar bears in blizzards in 3KB!

      --
      I watched C-beams glitter in the dark near the Tannhauser gate.
    3. Re:Compression - WOW! wait, I mean HOW? by GoRK · · Score: 3, Insightful

      I'm sure that they are not processing web traffic statistics, but that is an application that mirrors the data needs in a similar situation. I imagine many other applications have similar data needs. No doubt, the 3K is not re-expandable to 13TB, and probably neither is the 30GB.

      For instance, I might be generating 10GB of logs per day. To save disk space, the streams might be written out with gzip compression (it's write-only, after all), or I might rewrite the formats to conserve space - write the IP addresses in 4 bytes rather than writing out the full ASCII dotted quad, etc. Since it's text and the format is highly redundant, it compresses very well.. I might end up with a 50-500MB log file depending on how things are done. At the end of the day, I could process the statistics and generate report/archive data of several KB that retains all the important data I want to keep from the logs.

      ~GoRK

    4. Re:Compression - WOW! wait, I mean HOW? by AlXtreme · · Score: 1

      If they didn't, they just bought themselfs some really fancy hardware for their pr0n collection...

      --
      This sig is intentionally left blank
    5. Re:Compression - WOW! wait, I mean HOW? by Anonymous Coward · · Score: 0

      At the end of the day, I could process the statistics and generate report/archive data of several KB that retains all the important data I want to keep from the logs.

      That's not compression though - that's report generation.. the article specifically states compression..

      Compression of 100000:1 seems a little far fetched..

      I'm thinking that it's 3GB, not 3KB.

    6. Re:Compression - WOW! wait, I mean HOW? by GoRK · · Score: 1

      If 13TB can be compressed into 3GB then the system/code responsible for generating the original 13TB datastream should be reassesed as it is absolutely shit.

      Also, I never said anything about report generation, I said that I can process the logs and retain the important data that I need in a few KB - Most of the good web stats packages out there consolidate large log files into much smaller databases from which they can then run reports and get meaningful data from. Nobody ever said the 'compression' was lossless.

      Here's a better example, then. Take a simple voice recording: The original might be a 96KHz 24bit 8-channel file of someone saying "Hi" that is 50MB or so uncompressed. It could be compressed with some low bitrate codec to ~2400bps or so yeilding a file of 1-2KB. Finally I can 'compress' it even further and give you a 2 byte transcript of the file "Hi". From this, I could apply a simple text compression algorighm that uses 5 bits each to encode only letters A-Z and get the 2 byte string down to 10 bits: "HI" and you'd still have all the important content. Essentially this is 'compression' from 50MB -> 10bits or 41,943,040:1; however it's definately highly situation and highly data dependent -- and the article did not claim otherwise. 30GB->3KB is only 10,091,558:1, so my simple example is not that far fetched...

      ~GoRK

    7. Re:Compression - WOW! wait, I mean HOW? by GoRK · · Score: 1

      Correction: I did mention report generation; sorry.

    8. Re:Compression - WOW! wait, I mean HOW? by Gorbag · · Score: 1
      If 13TB can be compressed into 3GB then the system/code responsible for generating the original 13TB datastream should be reassesed as it is absolutely shit.
      Typically, real-time systems trade space for time. The 13TB is probably unprocessed raw data written pretty much as fast as the disks can keep up with it. Then a separate process can read and process the data ending up with a significant reduction. For instance, look at the size of the inputs to seti@home vs. the results sent back. Capice?
      --
      -- I speak only for myself
    9. Re:Compression - WOW! wait, I mean HOW? by GoRK · · Score: 1

      The price difference between what it costs to store 13TB of data vs 30GB of data is astronomical, especially when that extra 12.9TB is just temporary storage! The article says it's processed in real time to 30GB, so why write out the entire datastream to disk in the first place? It'd probably be cheaper to add a few machines with gobs of ram to cache the data then write out the 'processed' stream than to buy all that extra disk.

      I'm sure they have a reason for needing it though that this article fails to cover. I'd hope that almost anyone needing 13TB of storage is a smart person who evaluated their needs correctly.

      ~GoRK

    10. Re:Compression - WOW! wait, I mean HOW? by khb · · Score: 2, Interesting

      A long time ago, in a job far, far away, I had to deal with some radio astronomy data.

      The raw data was pretty much noise (quasar output).There were two data streams, from opposite sides of the earth, listening to the same noise source)

      The data (after extraction) was very, very small. I can't recall if it would have been on the order of Tb to Kb but it might well have been. We didn't have that much storage, so we had to compute (using custom hardware) in near real time.

      With appropriate temporary storage, much simpler (and probably more accurate) computational solutions are practical.

      It's possible they mean reduced not compressed (to use appropriate buzz words). Data reduction can be quite steep (e.g. count the number of LGM found in the current sample. Typically all your data reduces to 0 ;>).

    11. Re:Compression - WOW! wait, I mean HOW? by stux · · Score: 1

      Its a friggin massive radio-telescope. is it so hard to imagine needing 13TB to store the unprocessed data? :)

      perhaps they have two sets of 13TB disks and fill one while they process the other ;)

      --

      ---
      Live Long & Prosper \\//_
      CYA STUX =`B^) 'da Captain,
      Jedi & Last *-fytr
  2. Fix spelling "it's - its" by quake74 · · Score: 0, Troll

    Fix the spelling.

    1. Re:Fix spelling "it's - its" by quake74 · · Score: 1

      He he he, I didn't read that far ;) So much for the subscribers to catch spelling errors before stories are posted for the general public.

    2. Re:Fix spelling "it's - its" by TangoCharlie · · Score: 1

      You are quite correct. It should be its (no apostrophy). I definately think he should correct hi's spelling!

      --
      return 0; }
  3. mmmm....Xserve by GreatTeacherMusashi · · Score: 3, Funny

    Those things are soooo yummy, if I had a million dollars, some of it would go to an xserve array, just for fun, it'd be like the equivalent of some movie star buying like, some diamond coffee table or something "oh haha, and yeah that's my Xserve over there, oh no I don't really use it much, it's just to brighten the room up"

    --
    You win battles by knowing the enemy's timing, and using a timing which the enemy does not expect. Miyamoto Musashi
    1. Re:mmmm....Xserve by Sillypuddy · · Score: 1

      Bling Bling factor?
      http://www.urbandictionary.com/define.php ?term=Bli ng-Bling

      -joe

    2. Re:mmmm....Xserve by Anonymous Coward · · Score: 0

      Hell yeah... Once I make my first cool million $$, I plan to buy a few XServes. I figure with a coupe TB of storage I could encode all my CDs in FLAC and DVDs in MPEG-4. Then run gig-E throughout my house and have my own on-demand movie and music system.

    3. Re:mmmm....Xserve by Anonymous Coward · · Score: 0

      Personally, I'd buy 2 lian-li cases (I do like the brushed metal look) and fill those with drives and basic motherboards.

      Bet I could do it for about a third the cost of Apple's box too, with better cooling AND not supporting Apple. No warranty, but I'd be servicing it myself anyway :P

    4. Re:mmmm....Xserve by Anonymous Coward · · Score: 0

      No you wouldn't, as you simply cannot put together a system with the same features and price from off the shelf components. period.

    5. Re:mmmm....Xserve by zpok · · Score: 1

      You wouldn't be able to do what you say at the cost you state. Not with all the built in redundancy.

      Well, if you'd do this just for yourself, who cares. Then I can see the temptation to just go out and build the bastard. But I can't see institutions go for the "I come with the machine" deal ;-)

      --
      I think, therefore I am...I think.
    6. Re:mmmm....Xserve by Anonymous Coward · · Score: 0

      > AND not supporting Apple

      I don't see the problem in supporting a company that releases a good product. It's physical hardware, it's not like you can just will it into GNU freedom and make a dozen for yourself for free.

    7. Re:mmmm....Xserve by Anonymous Coward · · Score: 0

      I can top that: I've already done it. Sweet setup.

    8. Re:mmmm....Xserve by Anonymous Coward · · Score: 0

      Dude, learn how to use HTML! Don't post broken URLs with /. whitespace in them...

      Just make a link!

    9. Re:mmmm....Xserve by scoog · · Score: 1

      ... the ocean you hear? No my dear, we're fully three miles from the ocean here. You must be hearing my blessed. Xserve. Turn it off? Shall we go dancing my pet? ...

  4. Fiberchannel by emerrill · · Score: 1

    I wonder why apple would doubt that it would work (other then to try and get them to by Xservers with the Xserver RAIDs), they just use fiberchannel, which is a standard.

    1. Re:Fiberchannel by Anonymous Coward · · Score: 0

      It was most likely a "We only support the Xserver RAID when used with an XServe. As such, we cannot and do not offer any guarantee that it will work with Dells running Linux"

    2. Re:Fiberchannel by Anonymous Coward · · Score: 0

      Even more likely than that, it was, "Yes, we will be happy to sell you this Xserve RAID. Before we do, however, it's only fair to tell you that if you hook it up to a non-Apple system and have trouble, we won't even so much as take your call."

    3. Re:Fiberchannel by zpok · · Score: 1

      Which is a lot worse than "Yeah, it'll work, of course, nope, never tried it ourselves, but hey..."

      --
      I think, therefore I am...I think.
    4. Re:Fiberchannel by Anonymous Coward · · Score: 3, Informative

      You ever plugged a Fibre Channel RAID into a random Linux box? Your chances of getting it to work correctly without a lot of pain-in-the-assedness are about one in five.

    5. Re:Fiberchannel by weave · · Score: 2, Informative

      Interesting, we have 5 servers using qlogic 2200s. The default redhat kernel since 7.x works fine, autodetects the driver right out of the box. As long as the san zones are set up correctly, it'll find the disks with absolutely no problem.

    6. Re:Fiberchannel by Anonymous Coward · · Score: 0

      Interesting, we have 5 servers using qlogic 2200s. The default redhat kernel since 7.x works fine, autodetects the driver right out of the box.

      Why must you lie?

      I know this statement to be false. I installed a pair of Qlogic 2200s in a machine with Red Hat 7.1 on it, and they simply didn't show up. The disks attached were not visible.

      Why do Linux zealots find it necessary to lie?

    7. Re:Fiberchannel by Anonymous Coward · · Score: 0
      You're so full of shit. Did you do your zones correctly? Is there any type of access rules on your disks that you set up correctly for the host? When did Redhat go to lk 2.4? That makes a big difference. I know 7.0 was lk 2.2 which changes everything...
      Oct 15 08:35:24 hope kernel: (scsi): Found a QLA2200 @ bus 0, device 0x4, irq 22, iobase 0xcc00
      Oct 15 08:35:24 hope kernel: scsi(0): Allocated 4096 SRB(s)
      Oct 15 08:35:24 hope kernel: scsi(0): Configure NVRAM parameters...
      Oct 15 08:35:24 hope kernel: scsi(0): Verifying loaded RISC code...
      Oct 15 08:35:24 hope kernel: scsi(0): Verifying chip...
      Oct 15 08:35:24 hope kernel: scsi(0): LIP reset occurred
      Oct 15 08:35:24 hope kernel: scsi(3): Waiting for FW to complete...
      Oct 15 08:35:24 hope kernel: scsi(0): LIP occurred.
      Oct 15 08:35:24 hope kernel: scsi(0): LOOP UP detected
      Oct 15 08:35:24 hope kernel: scsi0: Topology - (FL_Port), Host Loop address 0x7d
      Oct 15 08:35:24 hope kernel: scsi-qla0-adapter-node=200000e06b03ba7d;
      Oct 15 08:35:24 hope kernel: scsi-qla0-adapter-port=210000e06b03ba7d;
      Oct 15 08:35:24 hope kernel: scsi-qla0-target-0=50060160010210df;
      Oct 15 08:35:24 hope kernel: scsi3 : QLogic QLA2200 PCI to Fibre Channel Host Adapter: bus 0 device 4 irq 22
      Oct 15 08:35:24 hope kernel: scsi(3:0:0:0): Enabled tagged queuing, queue depth 16.
      Oct 15 08:35:24 hope kernel: Attached scsi disk sdb at scsi3, channel 0, id 0, lun 0
      Oct 15 08:35:24 hope kernel: SCSI device sdb: 556696064 512-byte hdwr sectors (285028 MB)
      Oct 15 08:35:24 hope kernel: sdb: sdb1 sdb2
      Oct 15 08:35:24 hope kernel: scsi(3): Waiting for FW to complete...
      Oct 15 08:35:24 hope kernel: scsi3: Topology - (FL_Port), Host Loop address 0x7d
      Oct 15 08:35:24 hope kernel: scsi3: Topology - (FL_Port), Host Loop address 0x7d
    8. Re:Fiberchannel by Anonymous Coward · · Score: 0

      Did you do your zones correctly?

      No zones involved. Just storage.

      Is there any type of access rules on your disks that you set up correctly for the host?

      Of course not. Just storage, you idiot.

      When did Redhat go to lk 2.4?

      I'm sure I wouldn't have the foggiest idea. Nor should I care.

      Oct 15 08:35:24 hope kernel: (scsi): Found a QLA2200 @ bus 0, device 0x4, irq 22, iobase 0xcc00

      Congratulations. You either hacked your system for weeks to get it to work, or you learned to fake a system log excerpt. Huzzahs all around.

  5. Xserve is cheap by joshsnow · · Score: 4, Informative

    This topic reminded me of an article in the print version of UK mag PC Pro by Jon Honeyball. He tells of one of his (consultancy) "clients" looking for a SAN/RAID solution from Dell and EMC, which would cost 120,000.

    Aparently, Apples offering was 2TB storage for 9000 - vs 80,000 for 2TB from Dell/EMC.

    In the article he says;The obvious question is whether you can put fibre channel cards into your Windows servers and connect them up too. Apple tells me this is possible The idea being to use the storage for SQL*Server databases and the like.

    I'm not surprised this is such a good solution for use with Linux.

    1. Re:Xserve is cheap by drsmithy · · Score: 1
      Aparently, Apples offering was 2TB storage for 9000 - vs 80,000 for 2TB from Dell/EMC.

      I would propose that a price difference of that magnitude would indicate the Dell/EMC product is "doing a bit more".

      If the Dell/EMC box is stacked full of 15k RPM SCSI disks and a gig of mirrored cache, for example, then it's going to cost more because it's going to have better performance and reliability.

    2. Re:Xserve is cheap by stux · · Score: 1

      But when all you want is 2TB of storage which can sustain 100MB/s and has a good service contract from a company who isn't going out of business tomorrow...

      then you can pay 9,000 or 90,000

      --

      ---
      Live Long & Prosper \\//_
      CYA STUX =`B^) 'da Captain,
      Jedi & Last *-fytr
  6. Xserve RAID in non-Apple environments by daveschroeder · · Score: 4, Informative

    There's a new community site called alienRAID.org that focuses on supporting Xserve RAID in non-Apple and mixed environments. It's likely to be helpful to people interested in deploying Xserve RAID in conjunction with other platforms.

  7. Cheapest redundant solution by goombah99 · · Score: 4, Insightful
    Before a bunch of silly gooses shout they can get the same storagein their generic Linux boxes please lets note that were comparing reliable redundant systems.

    the X raid has dual redundant power supplies, redundant fans, dual redundant raid controllers, dual redundant and DEDICATED processors, dual redundant ethernet connection, dual redundant fiber channel outputs. it has separate busses and controllers for each ATI hard disk, and the busses to the disks are high speed. all of the disks are hot swapable self contained pluggin units. and it all sits in 3U. (plus another U for whatever server is receiving the fiber channel). All the software on board is tuned to the task and other than the web admin, the box has no extraneous services.

    also the raid is Hardware raid 5,1,0 not software. other than a netapp at 25x the price, there's nothing that comes close.

    --
    Some drink at the fountain of knowledge. Others just gargle.
    1. Re:Cheapest redundant solution by Anonymous Coward · · Score: 2, Informative

      Actually, while there are multiple controllers and cache modules in the Xserve RAID, they are not redundant. Each controller and cache module is responsible for its own side of the enclosure. The power supplies, cooling modules, etc., are fully redundant. The RAID Admin utility (the java piece), however, can talk to both controllers through only one IP address, so in effect, the ethernet devices are sort of redundant. This is all in the PDF documentation on the website, FYI.

      (Still a good deal for the money compared to some others, in my biased opinion.)

      (Posted anonymously because I work for Apple.)

    2. Re:Cheapest redundant solution by pmz · · Score: 2, Informative

      also the raid is Hardware raid 5,1,0 not software. other than a netapp at 25x the price, there's nothing that comes close.

      The only "disadvantage" of Apple's product is each individual disk is relatively slow (specs say 7200 RPM vs. 10000 to 15000 RPM for other options). Other than that, Apple really has a killer product (at least on paper...I don't have one :(

    3. Re:Cheapest redundant solution by Creepy+Crawler · · Score: 3, Insightful

      I'd rather have a bulk of HD's at 7200RPM than at higher speeds. More RPM equates to more heat and easier failure.

      --
    4. Re:Cheapest redundant solution by Anonymous Coward · · Score: 0

      Don't forget the XServe RAID uses ATA drives. Far less reliable than their SCSI equivalents. a set of 10 or 15K RPM SCSI drives would outperform, and likely last longer than the crud Apple are using.

    5. Re:Cheapest redundant solution by johnpaul191 · · Score: 1
      "crud"

      a bold statement from an anonymous post.

    6. Re:Cheapest redundant solution by Anonymous Coward · · Score: 0
      "Before a bunch of silly gooses shout..."

      geese ;-)

  8. Kewl. Everything to make a slashdot article. by kabocox · · Score: 5, Funny

    It runs linux check.
    It has Apple. Check.
    It has a low price for server hardware. check.

    This article is hearby Slashdot approved.

    1. Re:Kewl. Everything to make a slashdot article. by Maserati · · Score: 1

      Then I'm posting the obligatory spelling flame.

      "hereby"

      --
      Veteran, Bermuda Triangle Expeditionary Force, 1992-1951
  9. Um, well "Duh!". by DAQ42 · · Score: 4, Interesting

    I've been booting my Linux boxes from my XServe RAID for months now. It's just a fibre array. And the XRAID Admin program is a straight Java applications so you can run it on any platform that supports Sun Java 1.4.1.

    Go figure.

    It's not rocket science. These devices have been engineered using standards of the industry.

    Doh!

    --
    Don't Ask Questions. I don't know the answers and even if I did I wouldn't tell you.
  10. Apple's Trojan Horse for I.T. Depts.? by Johnny+Mozzarella · · Score: 5, Insightful

    iTunes has been called a Trojan Horse that will slip into consumer PCs and perhaps persude them to buy an iPod or a Mac.

    Will the XServe RAID become the equivalent Trojan Horse that will slip into corporate data centers and lead to future purchases of XServes?

    I hope Apple doesn't ignore this opportunity but instead promotes the fact that the XServe RAID plays well with other systems.

    1. Re:Apple's Trojan Horse for I.T. Depts.? by GreatTeacherMusashi · · Score: 1

      as much as I like Apple's products for the most part, they got an F in "Carpe Diem 101"...they won't say a thing...

      --
      You win battles by knowing the enemy's timing, and using a timing which the enemy does not expect. Miyamoto Musashi
    2. Re:Apple's Trojan Horse for I.T. Depts.? by Anonymous Coward · · Score: 0
      Knowing Apple's track record, I expect they'll take a page from Microsoft's book (i.e. "It's not done till Lotus won't run").

      Apple will retool and make sure that nothing like this ever happens again.

  11. ide raid by lithandie · · Score: 1
    I have picked up two hardware raid systems (Tornado) from these guys for about $7K each for 2TB in RAID 5. They have worked great so far and I have not had one problem (knock on wood).

    I am aware that the apple product is theoretically a little better due to the fiber chanel over the LVD SCSI that I purchased but I think that for most (not all) people the LVD performance would be good enough.

    I personally would recomend these Tornado RAID systems to just about anyone needing a lot of storage for a small price tag.

  12. I doubt Apple was surprised this worked by Ffakr · · Score: 4, Informative

    Apple went out of its way to describe how splendedly xServe RAID ran on Solaris, Linux, and even 2K/XP boxes at WWDC this year.
    This is the same story they've been telling since xServe RAID was shipping (and probably before that since it was pre-announced very early).

    I'm pretty sure that any doubts were from the customers or the Lunuxworld writer and not Apple. They've been trying to sell this for linux boxes. If it was 'Apple' that had doubts, then those engineers need to return to the mothership for some more training.

    --

    I'm not feeling witty so bite me

    1. Re:I doubt Apple was surprised this worked by Anonymous Coward · · Score: 0

      Win Win

      Doesn't work? No, we didn't think so.

      Ah, so it works. (customer doesn't care since it works)

  13. Carless use of cAsE in abbreviations by jerde · · Score: 1

    The linked article quotes the tech center's director:

    "Also, at around 100Mbps read and write to disk, it is faster than our SCSI system which has an I/O of 50Mbps"

    Never mind the whole base-2/base-10 "mebibyte" business... why can't people learn that the case of the "b"/"B" is significant!? Clearly he meant megaBytes, not bits.

    By the way, Apple advertises the Xserve RAID's throughput in excess of 200MBps.

    - Peter

    --
    INsigNIFICANT
    1. Re:Carless use of cAsE in abbreviations by Ffakr · · Score: 1

      Actually that's what I thought first also,...

      100 Mbit is only 12.5 MB/sec... pretty crappy for an IDE disk, but they did say per drive.

      The xServe RAID has two banks of 7 drives. Each bank has a 2000 Mbit fibre channel connection. So, let's do the math.

      2 Gbit / 7 drives is a max bandwidth of around 286 Mbit per drive.
      They quoted 100 Mbit sustained per drive.

      100 MB/sec x 7 drives would be 700 MB/sec of sustained bandwidth (not to mention that IDE drives don't sustain 100MB/sec.
      700 MB/sec is 5.6Gbit/sec, far in excess of even both the xServe RAID's fibre channels.

      I'm guessing that they were saying that they were getting 12.5 MB/sec of sustained throughput per drive, or 87.5 MB/sec of sustained throughput to one side of the xServe RAID.

      I think they might of actually meant Mbit...
      but I'm surprised the xServe didn't perform somewhat better.. and I'm really surprised the SCSI drives they were using were even worse. Perhaps they were doing tons of small reads or writes instead of streaming huge files over.

      ffakr

      --

      I'm not feeling witty so bite me

  14. Three Questions by Llywelyn · · Score: 1

    Three things come immediately to mind:

    1) Would it be hot swappable? (Remember: don't build systems that won't fail, build systems that fail gracefully).

    2) What would it cost?

    3) Is the fail rate so appreciably different that it matters and so that it would make up for the increased capital cost?

    --
    Integrate Keynote and LaTeX
  15. XServe Performance is good by Anonymous Coward · · Score: 0
  16. Something is wrong with the price by Luzumsuz+Lazim · · Score: 1
    Is it me, or is there really something wrong with the total cost given in the article?

    Here is my back of the envelope calculation:

    • $65,994 2.52 TB Xserve RAID (6x$10999)
      $45,117 into someone else's pocket
      --------
      $111,111 Subtotal
      -$11,111 Education and volume discount (10%)
      --------
      $100,000 Total
    Where did I make the mistake?
    1. Re:Something is wrong with the price by Anonymous Coward · · Score: 1, Informative

      $100,000 Australian = $69,917 US

    2. Re:Something is wrong with the price by rjstanford · · Score: 1

      Add in controller costs. Figure about $12648 for each XServe RAID, with full cache memory, et cetera, and another $5999 for a dual proc XServe to manage it. Of course, for something that size I'd run it through the ADC program if eligible ($3500 to join, drops the combined price from $18647 to $15307 per 2.5tb).

      --
      You're special forces then? That's great! I just love your olympics!
  17. Apple Australia SE told us... by icekey21 · · Score: 2, Interesting

    During the conference held couple weeks back at Melbourne Uni, the guy from Apple Australia explained to us as to why Swinburne TAFE decided to use the XRAIDS. As it's the only one that supports IDE (He went on to say that the whole reason why RAID is there for is to reduce the costs, yet others use the more expensive SCSI drives) it was the cheapest way to transfer the data from Parkes to Hawthorn. They basically get a van, pack the HDs onto it once the data is collected, and drive back to Melbourne, which is still cheaper and faster then transferring all the data through existing network channels. The ultmiate connectionless network =)

    1. Re:Apple Australia SE told us... by Anonymous Coward · · Score: 0

      Imagine what would happen when the van wrecks.

  18. Why they would doubt by ynotds · · Score: 1

    It's most likely a dealing with a branch office of a branch office problem. Swinburne is in Melbourne, Australia, while Apple Australian's main office is in Sydney.

    Over the years Apple Australia has managed to hire a few people technically competent to think outside the square, but more often, after such people have moved on to bigger challenges, they are left with a team which operates almost entirely on received wisdom.

    --
    -- Our systemic servants do not good masters make.
  19. Re:Apple RAID? by mikey1134 · · Score: 1

    Well, I'm just guessing that you haven't been actually reading any of the posts so I'll explain. WE ARE TALKING ABOUT THE XSERVE RAID. This has nothing to do with Mac desktops and there performance. The XServe RAID is an external fibre channel array, not a desktop. And in this case it is a lot cheaper than the competition.

    --
    <gir voice> I love this sig... </gir voice>