I have an issue with the 'spontaneous' in there - do you mean a deliberate explosion releases a different amount of energy?:) Actually, you mean if it all goes off at the same time. But that's not true either, I believe - you'd get the same amount of energy, over a longer period of time, by slowly decomposing the TNT
Where else would it explode? Did you expect it to move through the atmosphere at incredible speed without being affected, so it could explode lower down? Asteroids do not choose which part of the Earth to hit - there's atmosphere everywhere, and hence that's what they hit. The atmosphere compresses to create a bubble of heat and the asteroid burns up - explodes. As for EMP - this was not a nuclear explosion (which releases part of its energy directly as EM radiation). It was a mechanical explosion. Heat. Now a huge amount of heat does create EM radiation - just like something can glow red, then white, it will also glow in X-ray and eventually gamma rays (IANAP). But if there was enough energy to make an EMP, there would also be enough surviving material to hit the surface. And also we'd all be dead.
sorry, no "ballistic trajectory". And that's a strong point. Russia still has a missile tracking system. We all still do. We didn't shut them off after the end of the Cold War. These systems do not recognize something moving at ~60kmph as a missile - because it is not possible that such a projectile originated on Earth. We'd have noticed the launch - and the launch would have killed more people than the hit. This is something that exploded on impact with the atmosphere - not where ballistic missiles explode, and totally destroying it if it was a ballistic missile. To put it another way, if the asteroid was made of plutonium, you'd have a tiny eetsy bit of a radiation problem which you can easily shrug off. The mechanical result would have been the same.
Short story: There is an order-of-magnitude difference in speed, which makes the meteor impossible to track with missile-tracking systems, and these systems assume anything moving that fast is not a missile (because of the physical impossibility of it)
I often wonder where people get numbers. If you buy a smartphone, your cell phone bill will rise? How will they know you own a smartphone? I'm thoroughly bicurious. Confused. Sorry. Confused
Hmmm... I think perhaps that's the point. Everything is regulated, and has government intervention in it. Everything. Anything without any government in it is a purely private personal matter, and not done by corporations or places of business.
Umm... as a customer, I allow a store to take my money. I'm not forced to use Walmart, or Target, or Rami Levy. They're not a public institution. If I don't want to give them money, I either take only free samples or I don't enter the store This is ridiculous. As a customer I willingly hand over my money for a product. Why the hell should the government get involved here and impose a sales tax?
How would they gain an advantage when releasing their better product as open source? You can make a street so great it'll be better than any night club. But you still won't be making any money. "I started with the advantage. I kept the advantage the entire fight. And then I lost"
But but but... GPL! I don't know anything about copyright or the projects involved in the GNU/Linux OS or the people and politics behind it. I know 'GPL'! Wait. Are Linus and RMS different people?
In half a year, things may actually look rosy to the people who jump on the bandwagon. I say, let's ask again in about three, or five years. Sometimes, it takes time for an evil plan to come to fruition.
How is this comment a troll? There's no statement of a false fact here. There is no statement of any fact. There is an opinion, and it's not offensive or unrelated or anything that might be called a troll.
Slashdot does not have '-1 disagree'. '-1 troll' is definitely not the right substitute
7) Microsoft is starting to allow their own products like ASP.NET MVC to go FOSS.
Let me know if/when they ever finish that, until then it's like taking most of the mines out of a field.
I'd say it's like taking the mines out of the middle of the field, or just one edge. So there's still no way to cross it without getting blown up. (FOSS? Last I heard, Microsoft's released code was more like - don't modify this, this is not what we compile internally, it's for reading not for compiling, here have a lawyer cookie)
Yeah, he said Microsoft is making money from their monopoly So of course he meant they're not using the money for research and development. Just blow and hookers
If Nvidia or AMD did this to support their video cards, we would all be cheering about what a great thing this was. So why is it different for Microsoft?
Because Nvidia and AMD sell hardware. When Microsoft releases open source drivers for the hardware it sells, we all celebrate. Here, Microsoft releases open source code that provides support for their software. If you don't own that software, and the contribution is not intellectually fulfilling in any way, you gained nothing. The unwritten assumption here is that hardware is good for something, and you buy it to do that thing for you. You don't buy software so you don't have Microsoft's software platform to need support for. It's an assumption of complete lack of commercial software. It's probably valid in some academic circles
In other news, I agree that it's stupid to be angry at present Microsoft for things we felt back in the 90s. Microsoft is a corporation. Like people, it can change over time. And like crimes have a statue of limitations, the blame liability suffered by Microsoft must be finite. Otherwise they truly have no reason to become good - if they'll always be judged as evil.
people who clamp on GPL violations back it up with proof - i.e. showing what signs indicate that the violation occurred. They don't just tell companies to pay them royalties or else. Microsoft did not reveal which parts of the Linux kernel it found infringing, even though it has the code to read through. claims of GPL violations are based on evidence in the published binary: comparison with a known GPL binary, telltale signs of known code (symbol names) and sometimes the actual GPL itself quoted verbatim inside the binary These claims of violation come with hard data - which bytes are which values, and why that's bad. It's not a claim of 'you're violating the GPL because you copied GPL code. I won't tell you which GPL code you copied, or how I know, I'll just say that I'm the owner of that code'. That's insane.
To steal yourself away is to deny yourself from the current location When you steal a kiss, you deny someone else that kiss When you steal a look, you're looking at something before others do. When you steal an idea, you gain its advantages before the original creator
There never were other meanings to 'steal'
A woman has stolen my heart. Now my heart is not mine to command any more
If the object is not often moving and rubbing against other objects (mmmm... rubbing), then the coating need not be temporary. Dust-proof museum art. Spider-proof ceilings (seriously, this is all I can think of). I guess it's not very useful that way.
A stores half the data, B stores the other half, and C stores the parity. If a single one is down for maintenance, the data is readable.
That's an awful lot like storing parity on B for data on A. Since we've been comparing to RAID terms, dedicated parity storage is part of RAID 3 and RAID 4 - two levels which haven't been common in a long time (replaced with RAID 5 or RAID 6 - both featuring distributed parity).
Yes, RAID-5 or 6 would be much better - distributed parity - even when one of the providers is down, only some of the pieces will require recovery
In a mirrored arrangement of n mirrors, so long as 1 mirror is up, you can read the data. (n-1 mirrors can be down simultaneously.) "Obviously" there will be a resync after an outage.
So on. 7 providers. 5 hold data pieces and 2 hold parity
Or it's still better to just mirror them? cut the data into 7 pieces and write each piece on two providers
I see substantially diminishing returns after 2 providers, but your mileage may vary.
I intuitively see some truth in this, but why? * You don't gain any more speed
You already maxed out your downstream * You don't really gain more reliability
Reliability does not go higher and higher with more providers, because other components are still unreliable - i.e. you'll never reach 5 nines anyway. * You are "gaining" more complexity
Slower write performance?
I probably don't have to work within your justification framework.
No, I'm pretty sure you don't. I think your own seems to be just fine:)
I don't see a value in storing recovery information ("parity" in RAID parlance) on storage service B for the data on storage service A.
Agreed. Put like that, it seems stupid.
But what happens with 3 providers? A stores half the data, B stores the other half, and C stores the parity. If a single one is down for maintenance, the data is readable.
So on. 7 providers. 5 hold data pieces and 2 hold parity Or it's still better to just mirror them? cut the data into 7 pieces and write each piece on two providers
If you don't trust the provider to keep your data intact, don't use that provider.
That's either a ridiculous statement, or completely off-topic.
Neither, actually.
In a design like this, I assume that a storage resource - in this case, a cloud provider - will be either online, or offline. If they're offline, I need to work with a different copy of the data. Using a striping arrangement (or striping with parity) rather than a mirrored arrangement means there may not be another copy available.
So you agree that you planned for an outage - you planned for them to not keep your data intact - you didn't trust the provider. But you used it.
I think the statement is both Nobody trusts silicon or spindles, but we use them.
It's also offtopic because the question was not which provider to use. A question that many people here seem to be trying to answer for some reason
Fuller disclosure: * Storage is Amazon S3. No mention of other clouds. So it's just a worse version of S3 * Client is closed binary * Horrible 'acceptable use' policy and terms of use
oh yeah, it is closed source: "You must not reverse engineer or decompile the Software, nor attempt to do so, nor assist anyone else to do so"
So, um, yeah, they're data pirates waiting to kidnap your data. Have fun.
Oh and another thing its infinitely more secure to encrypt the data before "putting it up on your homemade mirror network" rather than as a process.
I'm not sure I understand - 'rather than as a process'. you mean rather than as part of the storage process? It's more secure to have the data already encrypted, before storing it.
For example, 99.99999999% of the data I "control" does not need to be encrypted. It just simply doesn't matter, even to a paranoid, although those know no rational limit....
OK, now you're just attacking him for wanting encryption.
Another example, lets say you were backing up a sql database of usernames/passwords for some site. The wrong way to do it is store the passwords in plain text and then encrypt the backup. Wrong for about a zillion (obvious?) reasons. If you have a decent system to hash and/or encrypt the data in the DB itself, thats much better, and no one can do anything with the encrypted data anyway. Or at least your database-level-backup script (as distinct from this project) can encrypt it for you (even if its just pipe mysqldump thru mcrypt and then into a file)
I agree, but still - you don't want the hashes to leak, either. (no matter the hash or salt, username+hash is way better than a web login interface, and if you have some knowledge about the user you might break it)
Why not encrypt everything as you store it, as well as (of course) keep salted hashes of passwords and not plaintext.
What about shared secrets and private keys? Should they be encrypted twice (before backup and at backup)?
Slashdot Mirror
3RC4!
I have an issue with the 'spontaneous' in there - do you mean a deliberate explosion releases a different amount of energy? :)
Actually, you mean if it all goes off at the same time. But that's not true either, I believe - you'd get the same amount of energy, over a longer period of time, by slowly decomposing the TNT
Where else would it explode?
Did you expect it to move through the atmosphere at incredible speed without being affected, so it could explode lower down? Asteroids do not choose which part of the Earth to hit - there's atmosphere everywhere, and hence that's what they hit. The atmosphere compresses to create a bubble of heat and the asteroid burns up - explodes.
As for EMP - this was not a nuclear explosion (which releases part of its energy directly as EM radiation). It was a mechanical explosion. Heat. Now a huge amount of heat does create EM radiation - just like something can glow red, then white, it will also glow in X-ray and eventually gamma rays (IANAP). But if there was enough energy to make an EMP, there would also be enough surviving material to hit the surface. And also we'd all be dead.
sorry, no "ballistic trajectory". And that's a strong point.
Russia still has a missile tracking system. We all still do. We didn't shut them off after the end of the Cold War.
These systems do not recognize something moving at ~60kmph as a missile - because it is not possible that such a projectile originated on Earth. We'd have noticed the launch - and the launch would have killed more people than the hit.
This is something that exploded on impact with the atmosphere - not where ballistic missiles explode, and totally destroying it if it was a ballistic missile.
To put it another way, if the asteroid was made of plutonium, you'd have a tiny eetsy bit of a radiation problem which you can easily shrug off. The mechanical result would have been the same.
Short story: There is an order-of-magnitude difference in speed, which makes the meteor impossible to track with missile-tracking systems, and these systems assume anything moving that fast is not a missile (because of the physical impossibility of it)
I often wonder where people get numbers.
If you buy a smartphone, your cell phone bill will rise? How will they know you own a smartphone?
I'm thoroughly bicurious. Confused. Sorry. Confused
The GPLv3 is a non starter in the enterprise world.
That must be why Android is such a commercial failure, eh?
Yes, that's exactly why there are no GPLv3 phones out there
GPLv2 phones rule the world. What's your point?
I was supposed to have a point?
The GPLv3 is a non starter in the enterprise world.
That must be why Android is such a commercial failure, eh?
yes, that must be it
Hmmm... I think perhaps that's the point.
Everything is regulated, and has government intervention in it. Everything.
Anything without any government in it is a purely private personal matter, and not done by corporations or places of business.
Umm... as a customer, I allow a store to take my money. I'm not forced to use Walmart, or Target, or Rami Levy. They're not a public institution. If I don't want to give them money, I either take only free samples or I don't enter the store
This is ridiculous. As a customer I willingly hand over my money for a product. Why the hell should the government get involved here and impose a sales tax?
How would they gain an advantage when releasing their better product as open source?
You can make a street so great it'll be better than any night club. But you still won't be making any money.
"I started with the advantage. I kept the advantage the entire fight. And then I lost"
But but but ... GPL! I don't know anything about copyright or the projects involved in the GNU/Linux OS or the people and politics behind it. I know 'GPL'!
Wait. Are Linus and RMS different people?
In half a year, things may actually look rosy to the people who jump on the bandwagon. I say, let's ask again in about three, or five years. Sometimes, it takes time for an evil plan to come to fruition.
How is this comment a troll?
There's no statement of a false fact here. There is no statement of any fact. There is an opinion, and it's not offensive or unrelated or anything that might be called a troll.
Slashdot does not have '-1 disagree'. '-1 troll' is definitely not the right substitute
7) Microsoft is starting to allow their own products like ASP.NET MVC to go FOSS.
Let me know if/when they ever finish that, until then it's like taking most of the mines out of a field.
I'd say it's like taking the mines out of the middle of the field, or just one edge. So there's still no way to cross it without getting blown up.
(FOSS? Last I heard, Microsoft's released code was more like - don't modify this, this is not what we compile internally, it's for reading not for compiling, here have a lawyer cookie)
Yeah, he said Microsoft is making money from their monopoly
So of course he meant they're not using the money for research and development. Just blow and hookers
The GPLv3 is a non starter in the enterprise world.
That must be why Android is such a commercial failure, eh?
Yes, that's exactly why there are no GPLv3 phones out there
If Nvidia or AMD did this to support their video cards, we would all be cheering about what a great thing this was. So why is it different for Microsoft?
Because Nvidia and AMD sell hardware.
When Microsoft releases open source drivers for the hardware it sells, we all celebrate.
Here, Microsoft releases open source code that provides support for their software. If you don't own that software, and the contribution is not intellectually fulfilling in any way, you gained nothing. The unwritten assumption here is that hardware is good for something, and you buy it to do that thing for you. You don't buy software so you don't have Microsoft's software platform to need support for. It's an assumption of complete lack of commercial software. It's probably valid in some academic circles
In other news, I agree that it's stupid to be angry at present Microsoft for things we felt back in the 90s. Microsoft is a corporation. Like people, it can change over time. And like crimes have a statue of limitations, the blame liability suffered by Microsoft must be finite. Otherwise they truly have no reason to become good - if they'll always be judged as evil.
people who clamp on GPL violations back it up with proof - i.e. showing what signs indicate that the violation occurred.
They don't just tell companies to pay them royalties or else.
Microsoft did not reveal which parts of the Linux kernel it found infringing, even though it has the code to read through.
claims of GPL violations are based on evidence in the published binary: comparison with a known GPL binary, telltale signs of known code (symbol names) and sometimes the actual GPL itself quoted verbatim inside the binary
These claims of violation come with hard data - which bytes are which values, and why that's bad. It's not a claim of 'you're violating the GPL because you copied GPL code. I won't tell you which GPL code you copied, or how I know, I'll just say that I'm the owner of that code'. That's insane.
To steal yourself away is to deny yourself from the current location
When you steal a kiss, you deny someone else that kiss
When you steal a look, you're looking at something before others do.
When you steal an idea, you gain its advantages before the original creator
There never were other meanings to 'steal'
A woman has stolen my heart. Now my heart is not mine to command any more
It's *our* microsoft
If the object is not often moving and rubbing against other objects (mmmm... rubbing), then the coating need not be temporary.
Dust-proof museum art.
Spider-proof ceilings (seriously, this is all I can think of). I guess it's not very useful that way.
The insides of windows?
Basically any object which is furniture.
A stores half the data, B stores the other half, and C stores the parity.
If a single one is down for maintenance, the data is readable.
That's an awful lot like storing parity on B for data on A.
Since we've been comparing to RAID terms, dedicated parity storage is part of RAID 3 and RAID 4 - two levels which haven't been common in a long time (replaced with RAID 5 or RAID 6 - both featuring distributed parity).
Yes, RAID-5 or 6 would be much better - distributed parity - even when one of the providers is down, only some of the pieces will require recovery
In a mirrored arrangement of n mirrors, so long as 1 mirror is up, you can read the data. (n-1 mirrors can be down simultaneously.) "Obviously" there will be a resync after an outage.
So on. 7 providers. 5 hold data pieces and 2 hold parity
Or it's still better to just mirror them? cut the data into 7 pieces and write each piece on two providers
I see substantially diminishing returns after 2 providers, but your mileage may vary.
I intuitively see some truth in this, but why?
* You don't gain any more speed
You already maxed out your downstream
* You don't really gain more reliability
Reliability does not go higher and higher with more providers, because other components are still unreliable - i.e. you'll never reach 5 nines anyway.
* You are "gaining" more complexity
Slower write performance?
I probably don't have to work within your justification framework.
No, I'm pretty sure you don't. I think your own seems to be just fine :)
I don't see a value in storing recovery information ("parity" in RAID parlance) on storage service B for the data on storage service A.
Agreed. Put like that, it seems stupid.
But what happens with 3 providers?
A stores half the data, B stores the other half, and C stores the parity.
If a single one is down for maintenance, the data is readable.
So on. 7 providers. 5 hold data pieces and 2 hold parity
Or it's still better to just mirror them? cut the data into 7 pieces and write each piece on two providers
If you don't trust the provider to keep your data intact, don't use that provider.
That's either a ridiculous statement, or completely off-topic.
Neither, actually.
In a design like this, I assume that a storage resource - in this case, a cloud provider - will be either online, or offline. If they're offline, I need to work with a different copy of the data. Using a striping arrangement (or striping with parity) rather than a mirrored arrangement means there may not be another copy available.
So you agree that you planned for an outage - you planned for them to not keep your data intact - you didn't trust the provider. But you used it.
I think the statement is both
Nobody trusts silicon or spindles, but we use them.
It's also offtopic because the question was not which provider to use. A question that many people here seem to be trying to answer for some reason
I'm still looking for the legal part where it says you're not allowed to connect to the Service with any kind of modified client.
Fuller disclosure:
* Storage is Amazon S3. No mention of other clouds. So it's just a worse version of S3
* Client is closed binary
* Horrible 'acceptable use' policy and terms of use
oh yeah, it is closed source:
"You must not reverse engineer or decompile the Software, nor attempt to do so, nor assist anyone else to do so"
So, um, yeah, they're data pirates waiting to kidnap your data. Have fun.
Oh and another thing its infinitely more secure to encrypt the data before "putting it up on your homemade mirror network" rather than as a process.
I'm not sure I understand - 'rather than as a process'. you mean rather than as part of the storage process? It's more secure to have the data already encrypted, before storing it.
For example, 99.99999999% of the data I "control" does not need to be encrypted. It just simply doesn't matter, even to a paranoid, although those know no rational limit....
OK, now you're just attacking him for wanting encryption.
Another example, lets say you were backing up a sql database of usernames/passwords for some site. The wrong way to do it is store the passwords in plain text and then encrypt the backup. Wrong for about a zillion (obvious?) reasons. If you have a decent system to hash and/or encrypt the data in the DB itself, thats much better, and no one can do anything with the encrypted data anyway. Or at least your database-level-backup script (as distinct from this project) can encrypt it for you (even if its just pipe mysqldump thru mcrypt and then into a file)
I agree, but still - you don't want the hashes to leak, either. (no matter the hash or salt, username+hash is way better than a web login interface, and if you have some knowledge about the user you might break it)
Why not encrypt everything as you store it, as well as (of course) keep salted hashes of passwords and not plaintext.
What about shared secrets and private keys? Should they be encrypted twice (before backup and at backup)?