RAID is the best backup



  • I started this at 3PM EST today. It might be done by 3AM. Re-syncing a 15TB RAID volume takes a little while.

    Edit: It doesn't take near as long as burning in a 48TB RAID volume though. Not even close.


  • sockdevs

    had a drobo, then the controller board went. called Drobo to see what i could do about getting my data back.

    their technicians response, and i'm not kidding, was:

    "Well we don't use standard RAID formats so your only option is to send the drives to us at your expense and pay us a restore fee, plus a fee for the new unit and we'll make a good faith effort to restore your data and ship the new unit back to you at your expense. Current estimated turn around time is 3 months from receipt of the old unit. Our data recovery fee in non-refundable if we fail to recover your data."

    copy pasted that from the chat log. yep. i said:

    "In that case you can keep my data and I thank the Goddess above that i kept backups of everything so that i'm not forced to fork over more money in the hope of maybe getting my data back.

    @accalia has disconnected from Chat"

    my replacement was a Qnap and i'm quite happy with it. and it's absolutely bog standard RAID so if something does go south anyone and their brother can provide a replacement RAID board for me to get my data back out of.


  • Grade A Premium Asshole

    @accalia said:

    "Well we don't use standard RAID formats so your only option is to send the drives to us at your expense and pay us a restore fee, plus a fee for the new unit and we'll make a good faith effort to restore your data and ship the new unit back to you at your expense. Current estimated turn around time is 3 months from receipt of the old unit. Our data recovery fee in non-refundable if we fail to recover your data."

    I do believe that their implementation is a variation on RAID4, combining JBOD with it. In theory, you should have been able to mount the drives individually under Linux and recover almost everything except files that spanned disks.


  • sockdevs

    theory isn't practice and that's a hell of an abusive strategy for data recovery.

    I had solid backups (including Amazon Glacier for the stuff that isn't music/movies (and therefore replaceable)) so i just cut my losses, one way or the other i was going to be out $1k and i'd rather be out it to a company that has a solid product and solid support, over one that's going to make me pay another $1k to get a replacement device (notice there was no talk of warranty, the device was 11mo wehn it went so would still be under most warrantees.) and maybe have my data recovered (they said good faith effort, nonrefundable)

    the Qnap is using absolutely standard RAID6 (which does mean i have to have drives of the same size in all the slots. I'm okay with that.) so i'm not tied to the vendor at all if something goes wrong

    (this is the NAS i bought. love the thing, even if the price has gone up $200 since i got it)


  • Grade A Premium Asshole

    Honestly, I would not use Drobos except that I get them for free. Their performance is on the slow side. For my purposes, I would prefer Netgear units. You are able to expand the RAID on theirs also and it is pretty much standard RAID4.

    The problem with most NAS units is that you are stuck with whatever configuration you set it up in. You cannot expand the RAID volume easily. With Netgear and Drobo units, you can pull one drive at a time and replace it with a larger drive, let it re-sync and expand your volume. Much more suitable for home environments. Most homes are unlikely to have some place to move more than 4TB of data to in order to expand (inexpensive external drive).

    If you have an 8-bay NAS unit running bog-standard RAID6 and you want to upgrade from 2TB drives you have to buy eight 4TB drives all at once, find some place to move ~12TB of data (so you are going to need three 4TB externals also), then spend the better part of a week moving data off, install all the new drives, and spend the better part of a week moving it back.

    Netgear and Drobo allow you to swap one drive at a time. No need for a temporary landing pad and all your data remains available for reads and writes through the process, just a little more slowly than usual.


  • sockdevs

    @Intercourse said:

    If you have an 8-bay NAS unit running bog-standard RAID6 and you want to upgrade from 2TB drives you have to buy eight 4TB drives all at once, find some place to move ~12TB of data (so you are going to need three 4TB externals also), then spend the better part of a week moving data off, install all the new drives, and spend the better part of a week moving it back.

    actually i had two options:

    1. Just fill it with 4TB drives from the beginning, i lost the source data and am going to have to restore from backups anyway
    2. use the 2TB drives i have lying around and buy 4TB drives one at a time until i have all 8 slots filled (the RAID will rebuild itself as i swap the drives one at a time) then expand the storage space. then expand the storage volume. (easy to do via web app or linux command line in the QNAPs)

    i went with #1 because i had 4 4TB already and just wanted to get that fiasco over. besides by the time i fill those it'll be time to replace with 8TB drives (or possibly a newer NAS to support the bigger drives.... depends on how long it takes to fill)


  • Grade A Premium Asshole

    I wasn't referring to your specific case. It was a generic statement. :smile:

    Regardless, with other options you could add a drive at a time and expand it that way.


  • sockdevs

    well if we're doing the general case...
    #RAID IS NOT BACKUP!

    ##You need a proper backup in place for when your RAID system falls over fatally, because if you don't it will.

    </rant>


  • Grade A Premium Asshole

    @accalia said:

    RAID IS NOT BACKUP!

    You need a proper backup in place for when your RAID system falls over fatally, because if you don't it will.

    I agree completely, but it is better than nothing for most home users, and nothing is what they had before.

    @PJH, could you move this thread of posts off in to another topic please? Or I think @boomzilla can do it also?


  • sockdevs

    @Intercourse said:

    @PJH, could you move this thread of posts off in to another topic please? Or I think @boomzilla can do it also?

    @abarker can too now.

    HULK MOVE POSTS! HULK SMASH THEM INTO NEW THREAD!

    err... sorry about that....



  • @Intercourse + @accalia

    Here you go.


  • sockdevs

    yay! Thank you.

    Also apparently these posts are now undread after the move. but?

    Note, I noticed the typo of undread before first post, i just thought it interesting enough not to correct.



  • Yeah, that's how it works.



  • Huh, so when posts are moved to a new topic they are marked as unread. I didn't think that posts were that strongly tied to what topic they were in (especially with all the moving that happens other places).

    Edit: and @accalia pointed it out before me. Guess I should have noticed it on meta.d though I follow it much less than here the moving is much much more common.


  • Grade A Premium Asshole

    Thank you very much sir.

    Now, back to the derailed but now on its own rails topic is not cluttering the other thread:

    RAID is a good component of backup for home users. Most home users do not need any of the file-serving functionality. They just need another copy of their files somewhere, which is all a backup is really. You have a copy on machine #1, and a copy on machine #2.

    But just using RAID in place of backups is a bad idea. But it is a much better idea than just having them on a desktop machine without RAID. It is a lesser of evils in that case. And for the users who don't need the file serving aspects of it, they could easily backup their machines to a NAS with Cobian Backup or a simple Robocopy script.



  • @Intercourse said:

    RAID is a good component of backup for home users. Most home users do not need any of the file-serving functionality. They just need another copy of their files somewhere, which is all a backup is really. You have a copy on machine #1, and a copy on machine #2.

    Well it depends.

    There are two prime usages for external RAID ( aka NAS, SAS, etc)

    1. Storage
    2. Backup

    Rule 1
    You can have "storage" without a "backup", but you can never have a "backup" without the "storage".

    Rule 2
    You don't serve files from backup

    Rule 3
    You make your Raid a small as possible, the bigger the capacity, the more moving parts - more point of failures.
    More failures Less chance of recovery.
    It is easier to rebuild 4tb of raid data then rebuild 12tb or 48tb
    ironically the rebuild process intensity can promote another disk failure

    Rule 4
    Get clean consistent regulated electricity. a UPS might help.
    some places you can cleaner electricity then others places. but also make sure that you don't make it dirty your self by connect a laser printer to the same line of power as your NAS.
    sudden power off are the main cause of data corruption, and power spikes can promote raid micro-controller errors.

    Rule 5
    Raid 5 is evil

    Rule 6
    Raid 6 is evil

    Rule 7
    make more then one backup for important stuff. make backup in more then one location.

    Rule 8
    there are always more rules



  • @Monarch said:

    There are two prime usages for external RAID ( aka NAS, SAS, etc)1. Storage2. Backup

    RAID isn't backup at all. If you accidentally delete or overwrite a file, it's gone from ALL of your RAID drives in a matter of milliseconds.

    @Monarch said:

    Rule 7 make more then one backup for important stuff. make backup in more then one location.

    That's good advice, but it works better if you tell people RAID isn't a backup at all. Which is true.


  • sockdevs

    @Monarch said:

    Rule 5Raid 5 is evil

    Rule 6 Raid 6 is evil

    [citations needed]



  • @blakeyrat said:

    RAID isn't backup at all. If you accidentally delete or overwrite a file, it's gone from ALL of your RAID drives in a matter of milliseconds.

    Exactly.

    The following is pushing things to make a point, but there's still a lot of truth to it: RAID is not for preventing data loss. It's for availability.


  • sockdevs

    @EvanED said:

    It's for availability.

    specifically for availability of data when drives fail. raid is resistant to drive failure by replicating your data among multiple drives. It affords a level of protection against a single drive failing, but it doesn't help for backup because all those copies of the data are in the same physical location. a fire/buildingcollapse, meteor strike/alien invasion can still wipe your data off the map because raid only protects against harddrive failures, and only so many of those at any given time.



  • @accalia said:

    but it doesn't help for backup because all those copies of the data are in the same physical location.

    AND HUMAN ERROR! The reason files are recovered from backup 95% of the time!


  • sockdevs

    @blakeyrat said:

    AND HUMAN ERROR! The reason files are recovered from backup 95% of the time!

    also this.

    rm -rf / is a dangerous tool.


  • I survived the hour long Uno hand

    I've been flirting with getting a NAS for two purposes:

    1. Storage of backups. I've had laptops stolen from my apartment before, and it makes me jumpy.
    2. Centralized storage of media. I've got a few external drives attached to a linux box that I'm using for this at the moment, but eventually that'll run out. With everyone in my household following different digital-media-purchase strategies (I buy from Amazon MP3 and Bandcamp, my roommate buys from iTunes, and my husband prefers CDs) it's best if we just dump the raw files into a centralized location to share music and movie rips.

    I'm torn between wanting the best and not wanting to spend a lot. The tech geek in my says "8TB! 16TB! As much space as can possibly be addressed! BACKUPS FOREVER!" while the fiscally responsible adult in me says "Eh.... how important is this really?"


  • Grade A Premium Asshole

    @Monarch said:

    Rule 3 You make your Raid a small as possible, the bigger the capacity, the more moving parts - more point of failures. More failures Less chance of recovery. It is easier to rebuild 4tb of raid data then rebuild 12tb or 48tbironically the rebuild process intensity can promote another disk failure

    There are exceptions to all of these. Speaking is absolutes only makes you incorrect in use cases that are outside of your sphere of influence.

    For instance, we manage several large storage RAIDs as offsite backup targets for our clients. The larger the volumes, the larger the dataset and the more efficiency we get out of dedups. It is easy to make these broad statements and make yourself seem smart, but there are exceptions to nearly all of these rules. (I agree with #2 and #4, but #4 works for anything electronic so it is still kind of silly to state.)

    @Monarch said:

    Rule 5Raid 5 is evil

    Rule 6 Raid 6 is evil

    Not necessarily. Mismanaged high-level RAIDs with shitty power connections are evil though. So if you think that RAID 5 and RAID 6 are evil, what does that leave? RAID 1? RAID 10? I hate losing 50% of my storage to redundancy overhead. It is entirely unnecessary unless you need the speed of RAID 10.

    ZFS claims to have fixed the "write hole" problem that comes up with uncommitted writes, but in the process introduced an entirely different set of issues that also cause your volume to shit itself if there is a power or hardware failure. Yeah, sure, you fixed one problem but put another problem in its place that has the exact same result. I am unimpressed.



  • My problem is that we don't have a good enough backup medium yet, in my mind. Tapes have had a mixed history of archival issues and are ridiculously expensive, spinning platters are guaranteed to die eventually because bearings can't last forever, and Flash? Well...we simply do not have the archival experience with floating-gate technology needed to evaluate it for backups; besides, I have lousy luck with dying USB Flash drives. Recordable optical media (WORM) might work, but RW optical media has not had a particularly good retention track record...

    And cloud backup? That's just making it someone else's problem. Tarsnap is the only service for that I have looked at seriously...

    Filed under: I wish we at least had some sort of magnetic removable disk...


  • sockdevs

    well if you ask me i'd say get something like this (8 bay because the 4bay is only $200 cheaper) and pick up 4 of these (in any size)

    that gives you 4 bays to expand into as your data grows and if you fill all the bays and still need more storage start replacing the drives with bigger ones, you can expand the storage once you've replaced them all with the new size and the array has rebuilt

    seems like a good balance between cost effective and room to grow.


  • Grade A Premium Asshole

    @Yamikuronue said:

    I'm torn between wanting the best and not wanting to spend a lot. The tech geek in my says "8TB! 16TB! As much space as can possibly be addressed! BACKUPS FOREVER!" while the fiscally responsible adult in me says "Eh.... how important is this really?"

    If you do get one, get something where you can add disks to down the line like Drobo or Netgear, then you do not have to buy a 16TB NAS right off the bat. Hell, if you have a little extra space you could always buy a diskless unit and shuck some external drives. Up to you though.

    I am sure @accalia will say differently, but we are all allowed to have differing opinions. ;)


  • sockdevs

    @Intercourse said:

    I am sure @accalia will say differently, but we are all allowed to have differing opinions.

    other than brand recommendations i think we're more or less spot on. ;-)


  • Grade A Premium Asshole

    @tarunik said:

    Filed under: I wish we at least had some sort of magnetic removable disk...

    ZIP disks!! ;)



  • @tarunik said:

    And cloud backup? That's just making it someone else's problem.

    So?

    Why not make it someone else's problems?

    Do you grow your own food? Synthesize your own engine oil? There's a billion things done for you every day that you've made into someone else's problem, and I don't see how backups is different.

    On the contrary: once you get a team of people and say "hey, you guys do backups and nothing but backups", they can generally do a much better job at backups than the team doing 47 different tasks. That's a thing called "specialization." It's pretty handy.

    The guys at Amazon running their Glacier backup service are pretty fucking good at it.


  • I survived the hour long Uno hand

    @accalia said:

    seems like a good balance between cost effective and room to grow.

    It also costs more than.... okay, most of the machines in my home right now. I bet my work machine costs more if I were to buy it myself without the benefit of a corporate discount that I'm sure they've negotiated.

    $973.99 + 4*$98.00 = $1365.99. That's more than any of the machines in my entertainment center; it's almost 40X as much as my favorite machine in that center. It's over twice as much as my laptop, and my husband's was cheaper. My desktop is really hard to price because I keep replacing parts, so it's less money than I've sunk into it but a lot more than it's worth.

    Then again, in terms of number of paychecks it'd take to save up, it's pretty reasonable. It's on par with the living room furniture I bought with my tax refund, or my nice bed. Still, it looks like a ridiculous amount to spend on hobbies >.> (I grew up poor*, can you tell?)

    poverty is relative; results may vary.



  • ?RAID == backup
    false
    ?RAID == FaultTolerance
    true
    ?RAID == FaultPrevention
    false
    ?RAID == FullFaultTolerance
    false
    

  • Grade A Premium Asshole

    ?RAID == HighAvailability
    true



  • @blakeyrat said:

    Why not make it someone else's problems?

    Because it creates an issue of 'who has control over your data?'; the service could go poof, or be served with, you know...papers.

    @blakeyrat said:

    On the contrary: once you get a team of people and say "hey, you guys do backups and nothing but backups", they can generally do a much better job at backups than the team doing 47 different tasks. That's a thing called "specialization." It's pretty handy.

    OTOH: specialization does help in this case, because it allows the service to amortize the cost of tape and other high-rel storage across its userbase.

    @blakeyrat said:

    The guys at Amazon running their Glacier backup service are pretty fucking good at it.

    Will look into this.

    @blakeyrat said:

    Do you grow your own food?

    Garden tomatoes are quite tasty! Same with peppers, for that matter.


  • sockdevs

    @Yamikuronue said:

    Then again, in terms of number of paychecks it'd take to save up, it's pretty reasonable. It's on par with the living room furniture I bought with my tax refund, or my nice bed. Still, it looks like a ridiculous amount to spend on hobbies >.> (I grew up poor*, can you tell?)

    well cost effective is relative too.

    I'd actually be very disappointed if you went out and bought something on our recommendation without doing your own research and decision making.

    you can get the cost of those discs down by getting consumer grade disks but that's asking for trouble. consumer disks try their best to hide errors from end user so if you use them by the time one fails there's a fair bet that the others are not far behind and might not make it through the rebuild cycle.

    if you do decide to go cheaper ther's always something like this:

    of course what you're losing is that the QNAP is itself a fairly powerful NAS that holds the RAID array and a full linux computer capable of doing all sorts of awesome stuf in addition to the raid things. the models with the i3 in them can even be media center PCs and do on the fly transcoding of full HD video (the model i linked is the atom processor which ios cheaper but not as powerful)

    you also lose the automatic PM features if you just go with the encosure. the Qnap will do regular checks of the drives to ensure they are functioning and will often give you drive failure warnings long before the drive actually goes giving you time to replace the drive and rebuild the array before anything bad happens, even if youo don't have the replacement on hand.

    but then the cost/benefit equation is really personal so it's up to you to say where you stand on that.


  • I survived the hour long Uno hand

    @accalia said:

    I'd actually be very disappointed if you went out and bought something on our recommendation without doing your own research and decision making.

    True, I was more reacting to the price point than the product. My usual recommendation-vetting process is to research reviews and read articles about what to look for, find something that looks decent, run it past my friend (A/N RED DIS IS U!!!!), dust off my ego as he tells me it's rubbish, and buy what he tells me :) he's yet to steer me wrong!



  • @tarunik said:

    Because it creates an issue of 'who has control over your data?';

    Encrypt yo' shit.

    @tarunik said:

    the service could go poof,

    True; but you gotta play the odds. I wager your business will crater decades before Amazon does.

    @tarunik said:

    or be served with, you know...papers.

    OH NOES! THE NSA BOOGEYMAN!!! Quick to the Slashdots!!!


  • Grade A Premium Asshole

    @blakeyrat said:

    True; but you gotta play the odds. I wager your business will crater decades before Amazon does.

    As far as Amazon, you likely have a point. They are unlikely to go under anytime soon, but it could happen. We have seen a few cloud providers go tits-up over the past 5 years. In one notable instance, clients were given 7 days to transfer all of their data out. The provider had petabytes of data and it would have taken months for everyone to get their data out. Tough shit, if you did not have another copy of it (and you should, but that is tengential to the point), you are fucked.

    @blakeyrat said:

    OH NOES! THE NSA BOOGEYMAN!!! Quick to the Slashdots!!!

    I consider NSA intrusiveness to be a really fucking major concern. It is an invasion of privacy and it has the potential to really hurt USA tech companies when competing in the global markets. So it is not just some Slashdot conspiracy.



  • @Intercourse said:

    I consider NSA intrusiveness to be a really fucking major concern. It is an invasion of privacy and it has the potential to really hurt USA tech companies when competing in the global markets. So it is not just some Slashdot conspiracy.

    I'm not saying it's not real, I'm saying it doesn't fucking matter and everybody who obsesses over it is literally crazy.

    Look, sometimes you find yourself in a group of people, and you look around the room and you realize: "everybody else here is certifiably crazy. I mean, seriously. These people need meds. You can't be this paranoid without some huge chemical imbalance. We're not talking 'ha ha' paranoid, we're talking 'about 3 weeks away from bombing a Federal building' paranoid here."



  • @blakeyrat said:

    Encrypt yo' shit.

    Rubber hoses, man. Although, if you have a good scheme (what tarsnap does comes to mind), that might be less of a concern b/c they actually have a clue about key management.

    @blakeyrat said:

    True; but you gotta play the odds. I wager your business will crater decades before Amazon does.

    Aye; Amazon cratering is a rather small risk in the grand scheme of things. Lesser-known providers, though...

    @Intercourse said:

    We have seen a few cloud providers go tits-up over the past 5 years. In one notable instance, clients were given 7 days to transfer all of their data out. The provider had petabytes of data and it would have taken months for everyone to get their data out. Tough shit, if you did not have another copy of it (and you should, but that is tengential to the point), you are fucked.

    @blakeyrat said:

    OH NOES! THE NSA BOOGEYMAN!!! Quick to the Slashdots!!!

    Hah! Is there anything in anyone's ToS that says 'we will notify you if we get subpoenaed for your data, and take your counsel's response into account when responding to the subpoena'?

    @Intercourse said:

    I consider NSA intrusiveness to be a really fucking major concern. It is an invasion of privacy and it has the potential to really hurt USA tech companies when competing in the global markets. So it is not just some Slashdot conspiracy.

    I consider the NSA's intrusiveness to be a major cybersecurity concern -- intentional weakening/backdoors that the blackhats are all too happy to take advantage of is not a possibility I want to deal with, given the size of the rubber hose the NSA has to use on those who don't play by their rules.

    Furthermore, their willingness to engage in active attacks against networks has shown no bound so far. What do you think the chance is that they'll stop and think before breaking into a PTC wayside server, for instance?



  • @tarunik said:

    Rubber hoses, man.

    ... what?

    @tarunik said:

    given the size of the rubber hose the NSA has to use

    ... the fuck are you talking about?

    @tarunik said:

    I consider the NSA's intrusiveness to be a major cybersecurity concern -- intentional weakening/backdoors that the blackhats are all too happy to take advantage of is not a possibility I want to deal with,

    I also don't want to deal with fictional bullshit from Tom Clancy novels.

    @tarunik said:

    Furthermore, their willingness to engage in active attacks against networks has shown no bound so far.

    We talking lower or upper bounds here? What "networks" have had "active attacks" against them?

    @tarunik said:

    What do you think the chance is that they'll stop and think before breaking into a PTC wayside server, for instance?

    I don't even know what a PTC wayside server is. And before you reply: I also don't care.


  • sockdevs

    @blakeyrat said:

    ... the fuck are you talking about?

    rubber hose cryptography.

    basically beating the shit out of someone until they give you they keys you want. either metaphorically or literally.



  • @boomzilla said:

    @Intercourse + @accalia

    Worst ship ever.

    @tarunik said:

    Filed under: I wish we at least had some sort of magnetic removable disk...

    I do have a stack of datasettes, if you want...

    @tarunik said:

    I consider the NSA's intrusiveness to be a major cybersecurity concern

    I've said it many times before, and I'll say it now - if you're publicly voicing concern about NSA peeking at your super secret data, it's almost certain that you're just information noise to them.

    The few people that should actually be worried about getting on NSA's hit list know better than to post on publicly accessible forums about how everyone's out to get them. And if they don't, well, they deserve it.


  • sockdevs

    @Maciejasjmj said:

    boomzilla:
    @Intercourse + @accalia

    Worst ship ever.

    indeed. i do not aprove.





  • @Maciejasjmj said:

    I've said it many times before, and I'll say it now - if you're publicly voicing concern about NSA peeking at your super secret data, it's almost certain that you're just information noise to them.

    The few people that should actually be worried about getting on NSA's hit list know better than to post on publicly accessible forums about how everyone's out to get them. And if they don't, well, they deserve it.

    Right; and what's worse is the issue is so muddled by paranoiacs and conspiracy theorists making-up those bad Tom Clancy plots and people who use terms like "rubber hose" to (apparently) mean beating a person(?) that, while I'm sure there's some actual valid concern here, I have no idea what that concern is.

    And any attempt to research it brings up rejected Tom Clancy plots.



  • @Intercourse said:

    There are exceptions to all of these.

    obviously, if there was only one truth in the world, the world would have been nothing but silence.

    Just a reminder your original statement which I used as a baseline to my response,
    referred to "Home Usage" / "Small User" use case

    And not to a company that provide a medium or large scale backup solution solutions.
    or any company that provide backup as a service. needs changes with scale.

    The common misconception that RAID == BACKUP can screw a "small user" big time.
    Many made that mistake and many will continue.

    As a small user if you already decided to go with Raid 5. and you give me the following options
    a. 1 x raid of 24TB
    b. 2 x raid of 12TB

    I would blindly prefer option b , you don't put all your eggs in one basket unless you don't mind a scrambled egg

    @Intercourse said:

    It is easy to make these broad statements and make yourself seem smart

    I do not pertain to look smart, after all we are as smart as our experiences.
    The smarts is the one whom learn from the mistakes of others. yet is not afraid to make its own mistakes.

    @Intercourse said:

    RAID 1? RAID 10? I hate losing 50% of my storage to redundancy overhead.

    my reasoning is use raid 5 or what ever you want. but at the end of the day. for a proper backup you will need redundancy.

    if you have a 12TB raid 5 on Location A
    and you mirror it to Location B to another 12TB raid
    beside the location difference you are going to lose 50% of you storage for redundancy overhead regardless.

    It is your choice if you want to increase the risk by spreading more data on less units.

    Bottom line
    RAID is like an insurance company it is all good until you need to make a claim.
    Even an insurance company have an insurance. make sure your RAID does too.


  • Discourse touched me in a no-no place

    @accalia said:

    [citations needed]

    Here's your citation: a customer I worked at was having problems with their application, which had a database on a RAID 5 array. They hired Progress (the company that makes the database product) to come out and tune it. They walked in, saw the RAID 5, and said "call us back when you're off RAID 5.)

    Another customer, our database was on a 30-spindle RAID 5 networked SAN. For reasons, I moved it to a pair of RAIDED drives--probably the mirrored one of 0/1, which I never CBA to remember is which--and nobody ever noticed a performance drop.

    The parity disk in a R5 array kills write performance if you're doing something that's heavily write-intensive. Home users will probably never tell the difference, though, nor would users who aren't using something like a database/app with a lot of transactions.

    This is well-known. You'll discover disk vendors routinely lie, though. "Our massive cache eliminates the slowdown due to parity." Sure, until you have to write enough to fill the cache.


  • sockdevs

    @blakeyrat said:

    Right; and what's worse is the issue is so muddled by paranoiacs and conspiracy theorists making-up those bad Tom Clancy plots and people who use terms like "rubber hose" to (apparently) mean beating a person(?) that, while I'm sure there's some actual valid concern here, I have no idea what that concern is.

    i'll let Randal have this one:

    the concern is:



  • @blakeyrat said:

    terms like "rubber hose" to (apparently) mean beating a person(?)

    Hey, we keep CompSci PG-13. "Mutilation with a crowbar", while probably more accurate, is not something you want to put in a research paper.

    Something to worry about if you're a military general holding access keys to a nuclear warhead. Something not to worry about if you're a basement dweller holding a pirated DVD of The Avengers and a few nudes of Jennifer Lawrence.


Log in to reply
 

Looks like your connection to What the Daily WTF? was lost, please wait while we try to reconnect.