Wow, Close call.kylef wrote:
SMART (a diagnostic system in my Western Digital hard drives) 'predicted' a crash. I was able to back up all my data and after about fifteen minutes the drive failed. Saved me!Flaming_Maniac wrote:
If they do keep them that long, the self diagnostic systems will detect the degradation of the disks so they can be replaced before any data loss. The assured death of a disk after a long period of time is a non-issue.
No self-diagnostics system can protect against a head crash, which is a common way for mechanical wear to present itself.Flaming_Maniac wrote:
People who are buying these high-end drives will be upgrading them before you can expect them to break. If they do keep them that long, the self diagnostic systems will detect the degradation of the disks so they can be replaced before any data loss. The assured death of a disk after a long period of time is a non-issue.mikkel wrote:
We aren't talking about upgrading to different disks. You can do that no matter how many disks you have, so it has no merit when trying establish the differences between having one and having multiple disks to store your data on. We're talking about a data loss scenario, and buying a new disk won't get the data on your old disk back. S.M.A.R.T. can't do anything about a sudden catastrophic failure, and even a "natural" disk death often manifests itself as a sudden failure, rather than a prolonged period of measurable wear.Flaming_Maniac wrote:
Chances are you will upgrade to a different disk or the S.M.A.R.T. system will pick it up before a natural disk death.
500GiB was a high-end disk three years ago. You don't see the people who bought 500GiB three years ago throwing those disks away today, so your assessment of operational life isn't accurate. You're also trying to tell me that the mechanical lifetime of a disk is irrelevant while discussing reliability with me. That is bordering on the absurd.
lol. We're talking about seriously premature disk crashes here and the consequences of hitting the one in a million chance. The chances of any one disk failing prematurely from some sort of manufacturer or user error increases directly with the number of drives you buy. Buying multiple drives is doing nothing more than saying you don't care about any individual drive so the increased risk of any one drive failing is fine, buying a single drive is saying you will play the (much lower) odds that you won't lose your only drive and all the data on it. I would rather back up my data and then have the simplicity and lower chance of any failure with a single drive, than plan for a failure with an array of drives.mikkel wrote:
No self-diagnostics system can protect against a head crash, which is a common way for mechanical wear to present itself.Flaming_Maniac wrote:
People who are buying these high-end drives will be upgrading them before you can expect them to break. If they do keep them that long, the self diagnostic systems will detect the degradation of the disks so they can be replaced before any data loss. The assured death of a disk after a long period of time is a non-issue.mikkel wrote:
We aren't talking about upgrading to different disks. You can do that no matter how many disks you have, so it has no merit when trying establish the differences between having one and having multiple disks to store your data on. We're talking about a data loss scenario, and buying a new disk won't get the data on your old disk back. S.M.A.R.T. can't do anything about a sudden catastrophic failure, and even a "natural" disk death often manifests itself as a sudden failure, rather than a prolonged period of measurable wear.
500GiB was a high-end disk three years ago. You don't see the people who bought 500GiB three years ago throwing those disks away today, so your assessment of operational life isn't accurate. You're also trying to tell me that the mechanical lifetime of a disk is irrelevant while discussing reliability with me. That is bordering on the absurd.
S.M.A.R.T. can look for signs of a head crash/malfunction, it monitors physical aspects of the hard drive. A hard drive's natural life is much longer than what most people would ever use it for.
But multiple drive solutions give you much more flexibility to resolve problems than single drive solutions. Disk accesses to secondary/storage drives tend to be far less frequent which means that the chances of the secondary drive failing are much reduced. This brings down the probability of secondary hard drive failures and means that the probability of catastrophic data loss is substantially lower. It makes far more sense to store sensitive data on a secondary drive in a two drive solution because the chance of failure is so much less.Flaming_Maniac wrote:
lol. We're talking about seriously premature disk crashes here and the consequences of hitting the one in a million chance. The chances of any one disk failing prematurely from some sort of manufacturer or user error increases directly with the number of drives you buy. Buying multiple drives is doing nothing more than saying you don't care about any individual drive so the increased risk of any one drive failing is fine, buying a single drive is saying you will play the (much lower) odds that you won't lose your only drive and all the data on it. I would rather back up my data and then have the simplicity and lower chance of any failure with a single drive, than plan for a failure with an array of drives.mikkel wrote:
No self-diagnostics system can protect against a head crash, which is a common way for mechanical wear to present itself.Flaming_Maniac wrote:
People who are buying these high-end drives will be upgrading them before you can expect them to break. If they do keep them that long, the self diagnostic systems will detect the degradation of the disks so they can be replaced before any data loss. The assured death of a disk after a long period of time is a non-issue.
500GiB was a high-end disk three years ago. You don't see the people who bought 500GiB three years ago throwing those disks away today, so your assessment of operational life isn't accurate. You're also trying to tell me that the mechanical lifetime of a disk is irrelevant while discussing reliability with me. That is bordering on the absurd.
S.M.A.R.T. can look for signs of a head crash/malfunction, it monitors physical aspects of the hard drive. A hard drive's natural life is much longer than what most people would ever use it for.
That I agree whole wholeheartedly with, and is basically what I have set up for an internal back up. Having a secondary drive in your system is not the same as buying two 500GB drives instead of one 1TB drive however.Bertster7 wrote:
But multiple drive solutions give you much more flexibility to resolve problems than single drive solutions. Disk accesses to secondary/storage drives tend to be far less frequent which means that the chances of the secondary drive failing are much reduced. This brings down the probability of secondary hard drive failures and means that the probability of catastrophic data loss is substantially lower. It makes far more sense to store sensitive data on a secondary drive in a two drive solution because the chance of failure is so much less.Flaming_Maniac wrote:
lol. We're talking about seriously premature disk crashes here and the consequences of hitting the one in a million chance. The chances of any one disk failing prematurely from some sort of manufacturer or user error increases directly with the number of drives you buy. Buying multiple drives is doing nothing more than saying you don't care about any individual drive so the increased risk of any one drive failing is fine, buying a single drive is saying you will play the (much lower) odds that you won't lose your only drive and all the data on it. I would rather back up my data and then have the simplicity and lower chance of any failure with a single drive, than plan for a failure with an array of drives.mikkel wrote:
No self-diagnostics system can protect against a head crash, which is a common way for mechanical wear to present itself.
500GiB was a high-end disk three years ago. You don't see the people who bought 500GiB three years ago throwing those disks away today, so your assessment of operational life isn't accurate. You're also trying to tell me that the mechanical lifetime of a disk is irrelevant while discussing reliability with me. That is bordering on the absurd.
S.M.A.R.T. can look for signs of a head crash/malfunction, it monitors physical aspects of the hard drive. A hard drive's natural life is much longer than what most people would ever use it for.
Not exactly the same, but similar. Many people will consider that sufficient storage for a system and only have that. If it's a choice between a system with 2 500GB drives or a system with a single 1TB drive, I know which I'd go for.Flaming_Maniac wrote:
That I agree whole wholeheartedly with, and is basically what I have set up for an internal back up. Having a secondary drive in your system is not the same as buying two 500GB drives instead of one 1TB drive however.Bertster7 wrote:
But multiple drive solutions give you much more flexibility to resolve problems than single drive solutions. Disk accesses to secondary/storage drives tend to be far less frequent which means that the chances of the secondary drive failing are much reduced. This brings down the probability of secondary hard drive failures and means that the probability of catastrophic data loss is substantially lower. It makes far more sense to store sensitive data on a secondary drive in a two drive solution because the chance of failure is so much less.Flaming_Maniac wrote:
lol. We're talking about seriously premature disk crashes here and the consequences of hitting the one in a million chance. The chances of any one disk failing prematurely from some sort of manufacturer or user error increases directly with the number of drives you buy. Buying multiple drives is doing nothing more than saying you don't care about any individual drive so the increased risk of any one drive failing is fine, buying a single drive is saying you will play the (much lower) odds that you won't lose your only drive and all the data on it. I would rather back up my data and then have the simplicity and lower chance of any failure with a single drive, than plan for a failure with an array of drives.
S.M.A.R.T. can look for signs of a head crash/malfunction, it monitors physical aspects of the hard drive. A hard drive's natural life is much longer than what most people would ever use it for.
No we aren't. You brought the disk replacement factor into this a few posts ago. That's a factor you threw in to help your case, but it simply isn't a factor that you can get away with counting on. We're talking about a data loss scenario, no matter how this may occur.Flaming_Maniac wrote:
lol. We're talking about seriously premature disk crashes here and the consequences of hitting the one in a million chance.mikkel wrote:
No self-diagnostics system can protect against a head crash, which is a common way for mechanical wear to present itself.Flaming_Maniac wrote:
People who are buying these high-end drives will be upgrading them before you can expect them to break. If they do keep them that long, the self diagnostic systems will detect the degradation of the disks so they can be replaced before any data loss. The assured death of a disk after a long period of time is a non-issue.
500GiB was a high-end disk three years ago. You don't see the people who bought 500GiB three years ago throwing those disks away today, so your assessment of operational life isn't accurate. You're also trying to tell me that the mechanical lifetime of a disk is irrelevant while discussing reliability with me. That is bordering on the absurd.
That's such an absurd twist for you to put on it. Using the same logic, you could say that buying a single disk is nothing more than saying that you don't care about all of your data enough to spread it out to prevent total loss when a disk does die. You can't defend your scenario by looking at it with the glass half full, and then attack my scenario by looking at it with the glass half empty. That's intellectually dishonest.Flaming_Maniac wrote:
The chances of any one disk failing prematurely from some sort of manufacturer or user error increases directly with the number of drives you buy. Buying multiple drives is doing nothing more than saying you don't care about any individual drive so the increased risk of any one drive failing is fine, buying a single drive is saying you will play the (much lower) odds that you won't lose your only drive and all the data on it. I would rather back up my data and then have the simplicity and lower chance of any failure with a single drive, than plan for a failure with an array of drives.
You keep talking about backups, but that's just so wholly irrelevant. This whole debate would be pointless if backup was in the picture, because we're talking about data loss, of which you would have none no matter how many drives you had, should the data be backed up.
S.M.A.R.T. can't find it if it occurs suddenly as a result of mechanical breakage. It's really that simple. It can't directly monitor the physical condition of a disk beyond temperature, because S.M.A.R.T. monitors in a way that assumes mechanical integrity. The best it can do is assume that one performance related symptom matches a known condition, but that requires the condition you ultimately suffer from to become gradually apparent.Flaming_Maniac wrote:
S.M.A.R.T. can look for signs of a head crash/malfunction, it monitors physical aspects of the hard drive. A hard drive's natural life is much longer than what most people would ever use it for.
There are too many ifs, ands and buts in your arguments, and it's more or less boiling down to saying that you can detect it using hardware features that a fraction of people even know exist, and even fewer monitor on a regular basis, or defending your stance using conditions that would make data loss impossible in either scenario.
In the end, I do this for a living, and the colleagues of mine that also do this for a living would agree. I've been fairly open to seeing things your way, but there are questions of mine that you haven't answered at all, and you've been dancing around the core issue by presenting conditions favourable to your argument, but irrelevant to our discussion. I'm not a close-minded person - I'm open to ideas. I think I've given this a bit too much, though, and it's probably better if we just stay here. If you have any closing comments, I'll of course read them. I'm not looking for the last word.
Last edited by mikkel (2008-07-05 23:41:48)
Yes, the important part is data loss, and if you know a drive is going bad, either because it's ancient or because some software tells you so, then you will go out and buy another disk before your disk goes bad if you value your data. If disk replacement is not a factor, it doesn't matter how your data is maintained, as long as it is maintained.mikkel wrote:
No we aren't. You brought the disk replacement factor into this a few posts ago. That's a factor you threw in to help your case, but it simply isn't a factor that you can get away with counting on. We're talking about a data loss scenario, no matter how this may occur.
I very plainly pointed out the negatives of one disk. You lose that disk, you lose everything. The flip side however is you increase the risk of losing some of your files. You still most likely have one working drive after a loss, but losing any files completely is a very negative scenario, and even if you do have them backed up, you still have to replace a drive.mikkel wrote:
That's such an absurd twist for you to put on it. Using the same logic, you could say that buying a single disk is nothing more than saying that you don't care about all of your data enough to spread it out to prevent total loss when a disk does die. You can't defend your scenario by looking at it with the glass half full, and then attack my scenario by looking at it with the glass half empty. That's intellectually dishonest.
You keep talking about backups, but that's just so wholly irrelevant. This whole debate would be pointless if backup was in the picture, because we're talking about data loss, of which you would have none no matter how many drives you had, should the data be backed up.
You brought the data loss point in here. The original argument stemmed only from the probabilities of drive failure, and while that usually goes hand in hand with data loss, they are not the same. In my opinion maintaining the lowest possible chance of any failure in your main computer (so you don't have to keep going back in and fucking with it) while maintaining a back up somewhere else in case of catastrophic failure is the best option.
So what are you doing, slinging your desktop around to lans? Giving is a good rap with a sledgehammer in the morning? Computers are mean to be used sitting on a table, in place, without a whole lot of moving around. Unless there is some serious user malfunction or the disk hasn't been broken in (not the 2 week RMA), the chances are it will die a slow and very detectable death.mikkel wrote:
S.M.A.R.T. can't find it if it occurs suddenly as a result of mechanical breakage. It's really that simple. It can't directly monitor the physical condition of a disk beyond temperature, because S.M.A.R.T. monitors in a way that assumes mechanical integrity. The best it can do is assume that one performance related symptom matches a known condition, but that requires the condition you ultimately suffer from to become gradually apparent.
The people that have a serious descion making process between two 500GB drives and one 1TB drive know about SMART.mikkel wrote:
There are too many ifs, ands and buts in your arguments, and it's more or less boiling down to saying that you can detect it using hardware features that a fraction of people even know exist, and even fewer monitor on a regular basis, or defending your stance using conditions that would make data loss impossible in either scenario.
These are the conditions that anyone seriously considering these drives are in. They do not want to lose any of their data if they can avoid it, they are going to use the full or close to full capacity, and they are well aware that making backups of important data in case of catastrophic loss is as important as saving word documents frequently. After all, hard drives dying is not always the fault of the hard drive. After all, you could lose an entire RAID array of who knows how many drives and the whole computer to one PSU gone bad, and in reality that has about the same liklihood as a drive dying. It's not "spinning it" or putting it in "favorable conditions" for my argument, because we don't live in a perfect world. If that's putting it in favorable conditions, then I guess that would be the best option to go with in the real world huh?
dancing? The only dancing I've been doing is condensing my responses to what I thought was the meat of your verbose arguments. If you think there is a serious gap in my responses please point them out.mikkel wrote:
In the end, I do this for a living, and the colleagues of mine that also do this for a living would agree. I've been fairly open to seeing things your way, but there are questions of mine that you haven't answered at all, and you've been dancing around the core issue by presenting conditions favourable to your argument, but irrelevant to our discussion. I'm not a close-minded person - I'm open to ideas. I think I've given this a bit too much, though, and it's probably better if we just stay here. If you have any closing comments, I'll of course read them. I'm not looking for the last word.
One drive. Simplicity. We're not trying to assure the continuation of the species of data by spreading it all out or something, we want to keep all of it.
^ Yes, my Samsung 300GB after ... I think this is coming up on three years Christmas - is dying. As a media drive it is quite important to me - and now there will even be a delay of up to 2-3 seconds in the middle of a song as it tries to read the rest. It's gone under heavy use but it hasn't failed yet! Hence my trust in Samsung.
I also do this for a living, and would have to agree.mikkel wrote:
In the end, I do this for a living, and the colleagues of mine that also do this for a living would agree.
That I wouldn't even begin to agree with. It's utter nonsense. You are making an argument based on people knowing what they are doing and then claim that they are going to use the full or close to full capacity of the drives - which anyone who does know what they are doing would have enough sense to avoid (primarily for maintenance and to keep access times low). You are making too many assumptions about user skill, without any real basis for doing so. This wouldn't be so bad except for the fact that you are pointing to a very specific level of user skill, not too high and not too low, to exactly fit your argument - which doesn't make any sense.Flaming_Maniac wrote:
These are the conditions that anyone seriously considering these drives are in. They do not want to lose any of their data if they can avoid it, they are going to use the full or close to full capacity, and they are well aware that making backups of important data in case of catastrophic loss is as important as saving word documents frequently. After all, hard drives dying is not always the fault of the hard drive. After all, you could lose an entire RAID array of who knows how many drives and the whole computer to one PSU gone bad, and in reality that has about the same liklihood as a drive dying. It's not "spinning it" or putting it in "favorable conditions" for my argument, because we don't live in a perfect world. If that's putting it in favorable conditions, then I guess that would be the best option to go with in the real world huh?
The chances of losing an entire system to one PSU gone bad are virtually nil. They are certainly far, far, far lower than the chances of unpredictable catastrophic data loss on a hard drive. In reality those chance are no where near the same.
That said, I do appreciate the point that by having 2 drives in the system you are essentially doubling the chances (you're typically not, because secondary drives usually fail far later than primary ones due to the number of accesses - but we'll overlook that) you will lose some data. But it is still obvious that a 2 drive solution is far and away the better option.
My only point here is that people are buying the drives because they actually need all of the space. That could be because they are a nub and are prepared to accept (or don't know) that using the complete disk will drastically lower read times for the files on the outer rims, or because they know they need the larger capacity drive so that they part they do use will keep the disks running in top form. All I'm doing is removing the cases of the people who are buying larger drives than they know they need, i.e. I only need about 400gb of space, so I'll get two drives and one can completely be a back up. Well of course that situation isn't fair to the single drive, because the 1000GB of capacity is not necessary. However an informed consumer needing a 1000GB of memory for only 400 gb of data because they don't want to fully use any one drive is the same situation as an ignoramus buying 10000GB of memory because they have 1000GB of data.Bertster7 wrote:
That I wouldn't even begin to agree with. It's utter nonsense. You are making an argument based on people knowing what they are doing and then claim that they are going to use the full or close to full capacity of the drives - which anyone who does know what they are doing would have enough sense to avoid (primarily for maintenance and to keep access times low). You are making too many assumptions about user skill, without any real basis for doing so. This wouldn't be so bad except for the fact that you are pointing to a very specific level of user skill, not too high and not too low, to exactly fit your argument - which doesn't make any sense.
That is a bit unclear, but I can't figure out any other way to say it. My point is user skill has no basis on the size of drive because you have to assume the consumer is buying as much memory as they need and not complete excess, whether they are filling that space with actual data or essentially sacrificing storage for performance.
Well, I've lost a hard drive (and an ethernet port) because of a PSU via lightning strike and I've lost a hard drive because I dropped it while trying to install it. It would be interesting if there was some statistical data out there, but I find it hard to believe that losing a PSU (and possibly other parts) is so many more times unlikely than losing a hard drive.Bertster7 wrote:
The chances of losing an entire system to one PSU gone bad are virtually nil. They are certainly far, far, far lower than the chances of unpredictable catastrophic data loss on a hard drive. In reality those chance are no where near the same.
I've already addressed and agreed to this. It is not fitting in a discussion of 500GBx2 vs. 1TBx1 because again we have to assume the user actually needs all the space for important data. If someone only cares about keeping half the data...well duh it's a no brainer.Bertster7 wrote:
(you're typically not, because secondary drives usually fail far later than primary ones due to the number of accesses - but we'll overlook that)
because? Based only on the ideal that you would rather increase the chances of keeping half your data than having decreased chances of keeping all your data?Bertster7 wrote:
That said, I do appreciate the point that by having 2 drives in the system you are essentially doubling the chances you will lose some data. But it is still obvious that a 2 drive solution is far and away the better option.
This thread is now officially considered, Hi-Jacked.
Yeah, I'll agree with that.Gooners wrote:
This thread is now officially considered, Hi-Jacked.
That's quite a big assumption.Flaming_Maniac wrote:
My only point here is that people are buying the drives because they actually need all of the space.Bertster7 wrote:
That I wouldn't even begin to agree with. It's utter nonsense. You are making an argument based on people knowing what they are doing and then claim that they are going to use the full or close to full capacity of the drives - which anyone who does know what they are doing would have enough sense to avoid (primarily for maintenance and to keep access times low). You are making too many assumptions about user skill, without any real basis for doing so. This wouldn't be so bad except for the fact that you are pointing to a very specific level of user skill, not too high and not too low, to exactly fit your argument - which doesn't make any sense.
Not really. Access times for the outer rim will remain constant. Access times for everything else will increase. The drive heads start reading from the outside edge and the drive spins faster at the outer edge, so access times are lower.Flaming_Maniac wrote:
That could be because they are a nub and are prepared to accept (or don't know) that using the complete disk will drastically lower read times for the files on the outer rims, or because they know they need the larger capacity drive so that they part they do use will keep the disks running in top form.
No it isn't. A 1TB drive for 400GB of data is a touch excessive, but not much. A 750GB drive (or 3x250GB drives in RAID5 - that's what I'd do) would be appropriate for a user requiring 400GB of data, a 500GB drive is clearly too small.Flaming_Maniac wrote:
All I'm doing is removing the cases of the people who are buying larger drives than they know they need, i.e. I only need about 400gb of space, so I'll get two drives and one can completely be a back up. Well of course that situation isn't fair to the single drive, because the 1000GB of capacity is not necessary. However an informed consumer needing a 1000GB of memory for only 400 gb of data because they don't want to fully use any one drive is the same situation as an ignoramus buying 10000GB of memory because they have 1000GB of data.
Well you are very much in a minority then. I've done a lot of work in hardware repairs and you very, very rarely see hard drive damage that has been anything to do with the PSU. Hard drive failure is very common as are PSU failures, but the two together is almost unheard of. I've seen the monthly statistics at places I've worked at, turning over about 1000 repairs a month - PSUs do not cause hard drives to fail, except in exceptional circumstances (which it sounds like you had - a lightning strike is hardly the normal way for hardware to fail).Flaming_Maniac wrote:
That is a bit unclear, but I can't figure out any other way to say it. My point is user skill has no basis on the size of drive because you have to assume the consumer is buying as much memory as they need and not complete excess, whether they are filling that space with actual data or essentially sacrificing storage for performance.Well, I've lost a hard drive (and an ethernet port) because of a PSU via lightning strike and I've lost a hard drive because I dropped it while trying to install it. It would be interesting if there was some statistical data out there, but I find it hard to believe that losing a PSU (and possibly other parts) is so many more times unlikely than losing a hard drive.Bertster7 wrote:
The chances of losing an entire system to one PSU gone bad are virtually nil. They are certainly far, far, far lower than the chances of unpredictable catastrophic data loss on a hard drive. In reality those chance are no where near the same.
It is still very relevant. Obviously they will have programs and their OS installed on the primary drive, reducing the secondary drives chances of failure massively. Therefore more important data can be stored on the secondary drive massively reducing the chances of any of the more important data being lost. Risk management. With more than one drive you have the option of selecting which data is most important - since very, very few users actually need all the data on their drives, or anywhere approaching that, it is a far more practical and realistic user model than the one you present.Flaming_Maniac wrote:
I've already addressed and agreed to this. It is not fitting in a discussion of 500GBx2 vs. 1TBx1 because again we have to assume the user actually needs all the space for important data. If someone only cares about keeping half the data...well duh it's a no brainer.Bertster7 wrote:
(you're typically not, because secondary drives usually fail far later than primary ones due to the number of accesses - but we'll overlook that)
For the reasons I just outlined about data prioritisation.Flaming_Maniac wrote:
because? Based only on the ideal that you would rather increase the chances of keeping half your data than having decreased chances of keeping all your data?Bertster7 wrote:
That said, I do appreciate the point that by having 2 drives in the system you are essentially doubling the chances you will lose some data. But it is still obvious that a 2 drive solution is far and away the better option.
Last edited by Bertster7 (2008-07-07 02:45:35)
This is becoming like a novel. :S
That's not cheap.
This is cheaper and better:
http://www.overclockers.co.uk/showprodu … =HD-053-SA
Last edited by Bertster7 (2008-07-07 02:47:47)
hhhmmmmmmmmBertster7 wrote:
That's not cheap.
This is cheaper and better:
http://www.overclockers.co.uk/showprodu … =HD-053-SA