This is insightful, but I don't think that including movie playback was where Flash lost its way. I think it was when they tried to make Flash into a "platform".
As you say, Flash started out as an animation plugin for vector graphics, and it was good at that. It became used more and more for advertising, which was annoying and often overkill; advertisers used Flash for things that would sometimes be handled more efficiently by a simple animated GIF. Today, it is generally good for 2 things: playing video and making casual games.
However, somewhere along the line, Adobe decided that it wasn't enough to handle content-creation, but they had to own a "platform". PDF stopped being a print-layout format, and suddenly you could build a whole little program into your PDF. Similarly, Flash stopped being an animation plugin and became something more of a development toolkit. When you look at Adobe Air, it becomes clear that Adobe wants you to build whole applications in Flash. Someone at Adobe is hoping that the future will see developers abandon other languages and development tools, and only Adobe will control the software industry.
Still, Flash as a vector animation plugin was doomed to obsolescence sooner or later. It's too simple a function for the world to be depending on a proprietary plugin.
I'd also be inclined to question: how much do you really need to destroy this data? What are you hiding? State secrets? Credit card information? Your porn-watching habits? Or your music collection?
A lot of times this stuff really boils down to "I don't really need to protect this data very thoroughly, but going through insane security measures makes me feel like I'm some kind of super-spy." In that case, it isn't security, it's ego-stroking.
If you can't count on your backups to be there when you need them, why have them at all? To me, that suggests a problem with the backup procedures more than a reason not to rely on having backups. But to each his own.
If I were a police officer wearing a bullet-proof vest, I still wouldn't want to run into the line of fire. That doesn't mean there's no reason to wear a bulletproof vest. Just because you've taken precautions doesn't mean it's smart to tempt fate. Especially so if you don't really know what you're doing.
when the files are measured in gigabytes, and the drives in terabytes, even "unlikely" errors start to become rather commonplace.... Personally, I have a script to verify the MD5 checksums of each file every 90 days on a rotating schedule... Occasionally I do remove a drive from the array to check the SMART status with a SATA/USB docking bridge..."
And this is part of my point: You know what you're doing, and you understand the risks. Fine, take the risk. If I'm giving advice, though, I'm generally giving it to people who don't understand these things (or why else give advice), so my advice has to be much more conservative.
Perhaps the insistence on "real" backups is part of the problem—if you save your data on tapes (for example),
What makes your online sync not a "real" backup isn't that it's online, but that it's always up-to-date. Mirroring isn't a "real" backup in my mind because if the original gets corrupted (which can be as big a problem as data loss) then that corruption gets passed over to the mirror. Backups keep old copies as you make updates. Online backups can do this.
For truly essential data, I would still have more confidence in multiple independent mirrors than in RAID1/4/5/6 arrays of the same capacity.
Well the redundancy in a RAID is not a backup either. You should be doing a backup in addition.
Yeah, I knew that what I was saying wasn't statistically accurate, but I didn't remember the right formula and I knew I wasn't going to take the time to figure it out for the sake of this argument. Especially not when most people here wouldn't blink at the mathematical inaccuracies, but would understand readily enough "If any of your 4 drives can die and ruin the whole array, then you have 4 chances for that to happen."
All that aside, based on years of practical IT experience, I still don't recommend that people rely on their backups. It's better to think of your backup as a last-ditch option rather than "something that will be there when you need it." The reason is pretty simple: most people won't actually do the work of checking their backups and verifying that the data is correct. People have a nasty habit of failing to notice that their backups are garbage until their data is gone and they're trying to restore.
If I were your IT guy, I'd also warn you that an rsync mirror is not really a backup, but you probably know that already and you're taking a calculated risk. That's all well and good as long as you know enough to calculate that risk. For most circumstances, though, I find the probability of a single disk failure is bad enough without increasing it by going with a RAID 0, especially if you're storing data that you feel you cannot afford to lose.
Well there's a question about what you're trying to accomplish, yes, and RAID 0 is faster. However, what you should overlook is that RAID 0 drastically increases your chances of losing the entire contents of your drive due to catastrophic disk failure.
When you get 2 hard drives together in a RAID 0, either one could die and cause total data loss, meaning you've doubled your chances of losing all the data on that RAID. There are consumer-grade RAIDs now offering 4 drives, which means that if you use a RAID 0 there, you've quadrupled your chances. This is a problem, because it's really not all that rare for a hard drive to fail.
So the point of "safe" RAID configurations is not just to increase safety over what a single drive provides (which is what RAID 1 does) but rather to mitigate the danger created by putting the drives in a RAID. A RAID 5 is much safer than a RAID 0, for example, because the chances of 2 drives failing at the same time is rather slim. However, there is bound to be a point where, if you have enough disks in a RAID 5 (I don't know how many, and it would depend on a couple factors), then it would be less safe than storing data on a single disk.
So... yeah, if you need a lot of speed and space for cheap, you're really confident in your backup solution, and you don't mind the downtime of needing to restore from backup if your RAID dies, then RAID 0 isn't a bad solution. But on the other hand, I don't recommend that people rely too heavily on their backups.
Whenever something like this comes up, lots of people go straight to thinking about which media will survive for 100 or 1000 years, but I don't think that's the right line of thinking. First, even if you buy a DVD that theoretically *can* last 100 years, that doesn't mean that your particular disk won't fail after 5 years. You have the possibility of improper storage, physical damage, manufacturing defects, or just... whatever. Random failure.
Mostly, treat it like other data. Create a backup plan, and make sure your backup plan is one that you'll actually stick to. People constantly come up with backup plans, do the backup once, forget, and then 2 years later their hard drive dies and they've lost everything in the last 2 years. In order to be a real backup, you need to keep backing up on a regular basis. The backup needs to be on a separate physical device, and you should periodically check the backup to make sure it's working, and that your backup copy hasn't been corrupted.
Ideally, you'll periodically send a backup offsite. This protects you from things like fires, floods, or other disasters where your entire house might be ruined. If you're writing stuff to a hard drive or DVD, periodically buy new backup media and send the old one to your parents or something. Or if the required storage is small enough, buy an online backup plan.
But the point here is, don't create 1 copy somewhere and expect that that copy on that medium will be around in 50 years. Instead, plan on moving the data to new media as newer/faster/larger media become available, and don't expect any medium to last more than 5 years or so. You'll get a new computer with a bigger hard drive, and you'll move the data to a new hard drive. The key here is to make sure that you don't lose your copy by accident, so keep good backups!
But seriously, is this really a concern when:
1. The Kindle Fire screen, battery, PCB, etc are all a single point of failure
Sure, but there's no way to get rid of that aside from buying multiple devices. But we're talking about a different kind of problem here: if all Kindle Fires use this service, then you're introducing a single point of failure for *all* of these devices. A single problem in Amazon's services could cause all Kindle Fires to be completely unable to access web pages, even when the device and the internet connection are fine. What's more, it's an unnecessary single point point of failure.
Didn't Opera do something similar, where they rendered the web pages on their own server and then passed it to their mobile browsers in an optimized form?
I believe that part of the problem is that it's often negotiated ahead of time. When they hire the CEO, there may be part of the contract negotiations that include huge payouts if they're fired.
The bigger question in my mind is, could these CEOs possibly be worth it? When you pay one of these yahoos $20 million, are you really getting $19 million more in value than if you hired the best $1 million CEO? With some of these CEOs, I doubt it.
Ok, now it's time for a bunch of people to complain about how snide and awful Mac users are, how they think that they're immune to security problems. We'll get a string of posts about how some study indicated that OSX was less secure than Windows, maybe some anecdotal evidence that some slashdotter knew a guy who was a Mac user, and he was an asshole and said something stupid about computers once.
When we've gotten enough of those, we'll see a backlash of posts rehashing old complaints about Windows and Linux, defending Macs.
My point wasn't about the particulars of human biology. My point is that when someone makes a prediction that something will happen in a set timeframe, and it doesn't happen in that timeframe, that doesn't suddenly mean that it will never happen.
Back in 2005, I remember watching a news show where an economist was complaining about the housing bubble, explaining that prices would eventually have to come down. They laughed at him. He explained his thinking, and I remember someone saying something to the effect of, "Yeah, but you predicted before that the bubble would burst in 2003, and it didn't. No, no. This is real economic growth that's just going to keep going. House prices will keep going up and up." That guy was wrong in 2003 and wrong in 2005, but things just happened later than he expected.
Well honestly the bigger issue is that you can't distribute H.264 content without a license.
But no, the fact that it's an open standard *does* matter. Maybe not in the short term of "What can I as a consumer do today?" but in a lot of non-obvious ways. Things like, "What are the practical concerns for using this as an archival format?" and "If I were developing a new platform that MPEG-LA didn't want to support, could I theoretically write an implementation myself?"
No, but I don't think I own other people's traffic over the internet either. Do you think that we should have privately owned infrastructure with no oversight? Should there be no rules about the practices of water, power, and sewage? Should private companies own the road with no rules about how they can restrict/control traffic?
Can GM buy the roads and make rules that only GM cars are allowed to drive on them? Can Amazon buy the roads and make FedEx pay extra tolls for transporting goods purchased from other online stores? Would that sort of thing be a good idea?
Internet access is a technological convenience, a service sold by private companies...
No, it's not. It's public telecommunication infrastructure. The fact that we allow private companies to built/maintain our vital infrastructure may be a mistake, but allowing them to control our infrastructure without oversight is moronic.
Honestly I feel like this is as much a problem as anything else. In my experience, contractors sort of lack "buy in". They have no skin in the game, so to speak. It's a little too easy for them to walk away without having really fixed your problem, and if anything it often ends up being in their interest to do a crappy job, because then you have to pay them to come back and fix everything that broke.
The 'theoretical limit' has always been 10 years away.
That doesn't mean the theoretical limit will always stay 10 years away. 20 years ago, the doctors said my father would be dead within a year. They were wrong, and he lived for another 20 years. However, if the doctors keep saying "he'll be dead within a year," then there will come a year when they'll be right. The fact that they were wrong 20 years ago does not mean that my father will live forever.
Yup, you don't even need soldiers coming back from war telling their horror stories. And better yet, if it gets truly automated, we won't even need to pay attention to whether we're killing who we meant to or achieving our goals. We can just set things loose in the world and assume that everything is going well.
Wait, so what are you confused about? VNC *is* full remote video and keyboard/mouse control. How else would you remotely install the OS unless the VNC session continued while the OS booted?
SSH just isn't better for the intended use here. It's worse. If it were just for BIOS control it would work, but it could mean learning complex commands and settings for each individual manufacturer and model. For a BIOS with a limited configuration options, a menu system is going to be easier and more intuitive than a command line.
But since the point is not just to allow remote configuration of the BIOS, but to allow total remote video/mouse/keyboard regardless of OS support, SSH is entirely inadequate.
Depending on the feature set, quality, and reliability, people may still want to buy the Lights-Out add-on cards. Either way, that's the way progress works sometimes. You're making money fixing problems, and then those problems go away. I don't think that Intel's, Dell's, or HP's business will be so hurt by this that it'll cause a huge hubbub.
Because what are you going to SSH into? The BIOS? Great, now you can change BIOS settings, and the whole system is completely useless once you boot your OS. Or are you going to SSH into your OS? Well first, that's no good for Windows, and second, we've already had remote logins on the OS level for a long time.
Sorry, but the value in something like this is to be able to see what's being displayed on the screen, regardless of what kind of output it is, and then to be able to use input devices (keyboard and mouse) for a total remote-access and remote-admin solution. VNC accomplishes that. SSH does not.
I would assume that this is something that is available in the BIOS, but that you can turn it off. The default should probably be for it to be turned off.
Slashdot Mirror
This is insightful, but I don't think that including movie playback was where Flash lost its way. I think it was when they tried to make Flash into a "platform".
As you say, Flash started out as an animation plugin for vector graphics, and it was good at that. It became used more and more for advertising, which was annoying and often overkill; advertisers used Flash for things that would sometimes be handled more efficiently by a simple animated GIF. Today, it is generally good for 2 things: playing video and making casual games.
However, somewhere along the line, Adobe decided that it wasn't enough to handle content-creation, but they had to own a "platform". PDF stopped being a print-layout format, and suddenly you could build a whole little program into your PDF. Similarly, Flash stopped being an animation plugin and became something more of a development toolkit. When you look at Adobe Air, it becomes clear that Adobe wants you to build whole applications in Flash. Someone at Adobe is hoping that the future will see developers abandon other languages and development tools, and only Adobe will control the software industry.
Still, Flash as a vector animation plugin was doomed to obsolescence sooner or later. It's too simple a function for the world to be depending on a proprietary plugin.
I'd also be inclined to question: how much do you really need to destroy this data? What are you hiding? State secrets? Credit card information? Your porn-watching habits? Or your music collection?
A lot of times this stuff really boils down to "I don't really need to protect this data very thoroughly, but going through insane security measures makes me feel like I'm some kind of super-spy." In that case, it isn't security, it's ego-stroking.
If you can't count on your backups to be there when you need them, why have them at all? To me, that suggests a problem with the backup procedures more than a reason not to rely on having backups. But to each his own.
If I were a police officer wearing a bullet-proof vest, I still wouldn't want to run into the line of fire. That doesn't mean there's no reason to wear a bulletproof vest. Just because you've taken precautions doesn't mean it's smart to tempt fate. Especially so if you don't really know what you're doing.
when the files are measured in gigabytes, and the drives in terabytes, even "unlikely" errors start to become rather commonplace.... Personally, I have a script to verify the MD5 checksums of each file every 90 days on a rotating schedule... Occasionally I do remove a drive from the array to check the SMART status with a SATA/USB docking bridge..."
And this is part of my point: You know what you're doing, and you understand the risks. Fine, take the risk. If I'm giving advice, though, I'm generally giving it to people who don't understand these things (or why else give advice), so my advice has to be much more conservative.
Perhaps the insistence on "real" backups is part of the problem—if you save your data on tapes (for example),
What makes your online sync not a "real" backup isn't that it's online, but that it's always up-to-date. Mirroring isn't a "real" backup in my mind because if the original gets corrupted (which can be as big a problem as data loss) then that corruption gets passed over to the mirror. Backups keep old copies as you make updates. Online backups can do this.
For truly essential data, I would still have more confidence in multiple independent mirrors than in RAID1/4/5/6 arrays of the same capacity.
Well the redundancy in a RAID is not a backup either. You should be doing a backup in addition.
Yeah, I knew that what I was saying wasn't statistically accurate, but I didn't remember the right formula and I knew I wasn't going to take the time to figure it out for the sake of this argument. Especially not when most people here wouldn't blink at the mathematical inaccuracies, but would understand readily enough "If any of your 4 drives can die and ruin the whole array, then you have 4 chances for that to happen."
All that aside, based on years of practical IT experience, I still don't recommend that people rely on their backups. It's better to think of your backup as a last-ditch option rather than "something that will be there when you need it." The reason is pretty simple: most people won't actually do the work of checking their backups and verifying that the data is correct. People have a nasty habit of failing to notice that their backups are garbage until their data is gone and they're trying to restore.
If I were your IT guy, I'd also warn you that an rsync mirror is not really a backup, but you probably know that already and you're taking a calculated risk. That's all well and good as long as you know enough to calculate that risk. For most circumstances, though, I find the probability of a single disk failure is bad enough without increasing it by going with a RAID 0, especially if you're storing data that you feel you cannot afford to lose.
Well there's a question about what you're trying to accomplish, yes, and RAID 0 is faster. However, what you should overlook is that RAID 0 drastically increases your chances of losing the entire contents of your drive due to catastrophic disk failure.
When you get 2 hard drives together in a RAID 0, either one could die and cause total data loss, meaning you've doubled your chances of losing all the data on that RAID. There are consumer-grade RAIDs now offering 4 drives, which means that if you use a RAID 0 there, you've quadrupled your chances. This is a problem, because it's really not all that rare for a hard drive to fail.
So the point of "safe" RAID configurations is not just to increase safety over what a single drive provides (which is what RAID 1 does) but rather to mitigate the danger created by putting the drives in a RAID. A RAID 5 is much safer than a RAID 0, for example, because the chances of 2 drives failing at the same time is rather slim. However, there is bound to be a point where, if you have enough disks in a RAID 5 (I don't know how many, and it would depend on a couple factors), then it would be less safe than storing data on a single disk.
So... yeah, if you need a lot of speed and space for cheap, you're really confident in your backup solution, and you don't mind the downtime of needing to restore from backup if your RAID dies, then RAID 0 isn't a bad solution. But on the other hand, I don't recommend that people rely too heavily on their backups.
I invested in a NAS Drive which has Raid 0
FYI, RAID 0 isn't very safe.
Whenever something like this comes up, lots of people go straight to thinking about which media will survive for 100 or 1000 years, but I don't think that's the right line of thinking. First, even if you buy a DVD that theoretically *can* last 100 years, that doesn't mean that your particular disk won't fail after 5 years. You have the possibility of improper storage, physical damage, manufacturing defects, or just... whatever. Random failure.
Mostly, treat it like other data. Create a backup plan, and make sure your backup plan is one that you'll actually stick to. People constantly come up with backup plans, do the backup once, forget, and then 2 years later their hard drive dies and they've lost everything in the last 2 years. In order to be a real backup, you need to keep backing up on a regular basis. The backup needs to be on a separate physical device, and you should periodically check the backup to make sure it's working, and that your backup copy hasn't been corrupted.
Ideally, you'll periodically send a backup offsite. This protects you from things like fires, floods, or other disasters where your entire house might be ruined. If you're writing stuff to a hard drive or DVD, periodically buy new backup media and send the old one to your parents or something. Or if the required storage is small enough, buy an online backup plan.
But the point here is, don't create 1 copy somewhere and expect that that copy on that medium will be around in 50 years. Instead, plan on moving the data to new media as newer/faster/larger media become available, and don't expect any medium to last more than 5 years or so. You'll get a new computer with a bigger hard drive, and you'll move the data to a new hard drive. The key here is to make sure that you don't lose your copy by accident, so keep good backups!
But seriously, is this really a concern when: 1. The Kindle Fire screen, battery, PCB, etc are all a single point of failure
Sure, but there's no way to get rid of that aside from buying multiple devices. But we're talking about a different kind of problem here: if all Kindle Fires use this service, then you're introducing a single point of failure for *all* of these devices. A single problem in Amazon's services could cause all Kindle Fires to be completely unable to access web pages, even when the device and the internet connection are fine. What's more, it's an unnecessary single point point of failure.
Sounds like we have an MBA in the house.
Didn't Opera do something similar, where they rendered the web pages on their own server and then passed it to their mobile browsers in an optimized form?
I believe that part of the problem is that it's often negotiated ahead of time. When they hire the CEO, there may be part of the contract negotiations that include huge payouts if they're fired.
The bigger question in my mind is, could these CEOs possibly be worth it? When you pay one of these yahoos $20 million, are you really getting $19 million more in value than if you hired the best $1 million CEO? With some of these CEOs, I doubt it.
Ok, now it's time for a bunch of people to complain about how snide and awful Mac users are, how they think that they're immune to security problems. We'll get a string of posts about how some study indicated that OSX was less secure than Windows, maybe some anecdotal evidence that some slashdotter knew a guy who was a Mac user, and he was an asshole and said something stupid about computers once.
When we've gotten enough of those, we'll see a backlash of posts rehashing old complaints about Windows and Linux, defending Macs.
You *can* write your own implementation. You'll just have to pay a licensing fee.
My point wasn't about the particulars of human biology. My point is that when someone makes a prediction that something will happen in a set timeframe, and it doesn't happen in that timeframe, that doesn't suddenly mean that it will never happen.
Back in 2005, I remember watching a news show where an economist was complaining about the housing bubble, explaining that prices would eventually have to come down. They laughed at him. He explained his thinking, and I remember someone saying something to the effect of, "Yeah, but you predicted before that the bubble would burst in 2003, and it didn't. No, no. This is real economic growth that's just going to keep going. House prices will keep going up and up." That guy was wrong in 2003 and wrong in 2005, but things just happened later than he expected.
Well honestly the bigger issue is that you can't distribute H.264 content without a license.
But no, the fact that it's an open standard *does* matter. Maybe not in the short term of "What can I as a consumer do today?" but in a lot of non-obvious ways. Things like, "What are the practical concerns for using this as an archival format?" and "If I were developing a new platform that MPEG-LA didn't want to support, could I theoretically write an implementation myself?"
No, but I don't think I own other people's traffic over the internet either. Do you think that we should have privately owned infrastructure with no oversight? Should there be no rules about the practices of water, power, and sewage? Should private companies own the road with no rules about how they can restrict/control traffic?
Can GM buy the roads and make rules that only GM cars are allowed to drive on them? Can Amazon buy the roads and make FedEx pay extra tolls for transporting goods purchased from other online stores? Would that sort of thing be a good idea?
Internet access is a technological convenience, a service sold by private companies...
No, it's not. It's public telecommunication infrastructure. The fact that we allow private companies to built/maintain our vital infrastructure may be a mistake, but allowing them to control our infrastructure without oversight is moronic.
Do you think we should let FedEx own the roads?
an accountability nightmare.
Honestly I feel like this is as much a problem as anything else. In my experience, contractors sort of lack "buy in". They have no skin in the game, so to speak. It's a little too easy for them to walk away without having really fixed your problem, and if anything it often ends up being in their interest to do a crappy job, because then you have to pay them to come back and fix everything that broke.
The 'theoretical limit' has always been 10 years away.
That doesn't mean the theoretical limit will always stay 10 years away. 20 years ago, the doctors said my father would be dead within a year. They were wrong, and he lived for another 20 years. However, if the doctors keep saying "he'll be dead within a year," then there will come a year when they'll be right. The fact that they were wrong 20 years ago does not mean that my father will live forever.
Nope. They probably tried to hire Carly Fiorina first, but couldn't get her.
Yup, you don't even need soldiers coming back from war telling their horror stories. And better yet, if it gets truly automated, we won't even need to pay attention to whether we're killing who we meant to or achieving our goals. We can just set things loose in the world and assume that everything is going well.
Wait, so what are you confused about? VNC *is* full remote video and keyboard/mouse control. How else would you remotely install the OS unless the VNC session continued while the OS booted?
SSH just isn't better for the intended use here. It's worse. If it were just for BIOS control it would work, but it could mean learning complex commands and settings for each individual manufacturer and model. For a BIOS with a limited configuration options, a menu system is going to be easier and more intuitive than a command line.
But since the point is not just to allow remote configuration of the BIOS, but to allow total remote video/mouse/keyboard regardless of OS support, SSH is entirely inadequate.
Depending on the feature set, quality, and reliability, people may still want to buy the Lights-Out add-on cards. Either way, that's the way progress works sometimes. You're making money fixing problems, and then those problems go away. I don't think that Intel's, Dell's, or HP's business will be so hurt by this that it'll cause a huge hubbub.
Because what are you going to SSH into? The BIOS? Great, now you can change BIOS settings, and the whole system is completely useless once you boot your OS. Or are you going to SSH into your OS? Well first, that's no good for Windows, and second, we've already had remote logins on the OS level for a long time.
Sorry, but the value in something like this is to be able to see what's being displayed on the screen, regardless of what kind of output it is, and then to be able to use input devices (keyboard and mouse) for a total remote-access and remote-admin solution. VNC accomplishes that. SSH does not.
I would assume that this is something that is available in the BIOS, but that you can turn it off. The default should probably be for it to be turned off.