What address does an IPv4 host use to respond to an IPv6 host after the IPv4 pool is exhausted? It can't be IPv4 because there are none for the IPv6 host to be assigned and it can't be IPv6 because IPv4 does not understand IPv6.
DJB is suggesting something like an IPv4 plus with a compatible, routable address space. Meaning those IPv4 hosts instead of being upgraded to dual stack would be upgraded to single stack IPv4 plus, no network reconfiguration or renumbering required.
Every IPv4 configuration in the world would be a viable IPv4 plus configuration. Major corporations could switch to IPv4 plus without knowing what happened. Network admins could be blissfully ignorant of the expanded address space for years.
After this silent deployment had taken place during the natural refresh cycle, IPv4 plus addresses in the extended address space would be reachable by virtually everybody and could be used in a _single stack_ configuration without major headaches.
As it happens there was such a proposal before the IETF back in 1993 (aka "TP/IX", RFC1475), but the IETF rejected it because they wanted a cleaner, "reboot the world" design. Bad mistake.
Any "fix" to this would require changing how IPv4 anyway, which is WORSE than switching to IPv6.
A proper fix to IPv4 would have the major advantage of being configuration compatible. A large corporation could upgrade to "IPv4 plus" and not know that it happened. It would just work. No dual stacking, no renumbering, no network reconfiguration, let alone maintaining two independent configurations at once.
Once most of the organizations in the world had IPv4 plus compatible devices (without even knowing about it) addresses in the extended area could be used. That would be a _much_ cleaner transition plan than the reboot the world plan associated with IPv6. It could have happened a decade ago and most people would neither know nor care.
There is a trivial mapping between IPv4 and IPv6 addresses
Yes, but it is not supposed to be one that anyone is supposed to use, except as a temporary crutch. The IETF wants the entire existing network configuration of the world to go away. That includes all v6 mapped v4 addresses, network prefixes, and so on.
A more viable transition plan would have the IPv6 backbone route those prefixes forever, as opposed to the current plan, which to first approximation is "not at all" - not on the v6 backbone at any rate.
The author claims that IPv6 should have been designed as an extension to IPv4 so that IPv4 and IPv6 hosts can communicate with each other directly. This is fundamentally impossible.
On the contrary, if IPv6 address space was designed to be an long term extension of the IPv4 address, dual stacking would not be required. Network reconfiguration wouldn't be required either.
DJB recognizes that of course everyone will eventually have to get upgrades to handle the extra address space, or for anyone to practically use an address in the extended area. The difference is that the upgrade would be transparent, because no network reconfiguration would have to be done. All old IP addresses and routing prefixes would be preserved, on the new network, forever.
After a few years, everyone would have extended address space compatible software and hardware and it would just start working automatically with extended addresses without any user intervention required. Network administrators, in particular, would be required to do precisely nothing. That is a major advantage in the real world.
position #1 being the recurring myth that ISPs are subject to common carrier regulations.
The only reason why ISPs are not subject to common carrier regulations is because the FCC is wildly (and opportunistically) misreading the Communications Act, which defines "Telecommunications" as follows:
-The term ''telecommunications'' means the transmission, between or among points specified by the user, of information of the user's choosing, without change in the form or content of the information as sent and received
Sounds like Internet access service, no? In fact there has _never_ been a better example of telecommunications service than Internet access service.
So while you are technically right as a matter of current federal regulation (which can be changed/corrected at any time by the FCC) that Internet access service is not subject to Title II common carrier regulations, all those Slashdot users you so blithely accuse are arguably right as a matter of law. It is only the worst kind of opportunistic FCC legerdemain that keeps the law from being enforced as written.
Title II is far more sound as a basis for net neutrality regulations than whatever crazy legal theory the FCC has been promoting lately to regulate broadband access while simultaneously pretending that it is not a form of "telecommunications". And if they are forced into that position, that is precisely what the FCC will do - read the law the way it was written, objections from all the people who are horrified at the prospect notwithstanding.
First of all, it is technically open source, but the license the community edition uses means it cannot legally be used by businesses.
That is not technically "open source" by any common definition, most particularly not that of the Open Source Initiative, which prohibits discrimination against fields of endeavor.
Please clarify what you mean by forever. The IPv4 to IPv6 mapping is forever. Nobody is going to come and steal away that little 32 bit slice of the v6 address space.
No, the addresses will simply be useless as soon as the IPv4 net either ceases to exist or ceases to be routed to the location of the addressee. Strictly a stop gap measure. The IPv4 numbering plan is going away, and that means that every public facing server on the planet will have to be dual stacked (with two completely independent addresses) for at least a decade.
You can reach any v4 host from v6 using this mapping using NAT64. And you will be able to do that as for as long anyone cares to keep the v4 net alive together with a NAT64 gateway.
Aye, that's the rub.
No, there is not actually any way v6 could have been constructed in a different way that would have made the ISPs of the world adopt it sooner. They will adopt when forced to, no sooner, no matter how it works.
Not true. The scheme I described, would allow existing networks to keep their IP addresses and network numbering plan forever, because the new numbering plan would be a transparent expansion of the old numbering plan. Every existing IP address, network, and subnet would be an IP next generation address, so far as the user could tell. No need renumber anything. No new DNS entries, no dual stack, etc. Over five or ten years, everyone would have IP next generation hardware and software and not even know it, or need to do anything to put it into effect. That includes ISPs and network administrators of all stripes.
An ISP could literally transition to IP next generation without knowing the transition took place, because they would not have to allocate replacement addresses for anyone. There wouldn't be two parallel networks, two parallel backbones, two sets of DNS entries for everything and everything.
Instead the natural upgrade cycle would have brought everyone everywhere into IP next generation compliance in less than a decade without them even knowing about it. And then as old style IP addresses were depleted, the few stragglers remaining would actually become aware they needed some sort of router upgrade. Not new IP addresses, not a new numbering plan, not new DNS entries, just a router upgrade. Done.
The entire transition would mostly be visible only to programmers and hardware engineers. Network administrators would hardly have to care. The transition, instead of being a world wide ten year headache, would pass with hardly anyone knowing about it. That is what a conservative addressing expansion plan would look like. No attempt to reboot the world from scratch.
They did this. Except they added 12 octets in front of v4 and mapped existing v4 addresses to 0.0.0.0.0.0.0.0.0.0.0.0.x.x.x.x.
A clean transition would have to support a conformal mapping of the current IPv4 address space forever. IPv6 isn't designed to do that, so the IPv4 mapping is useless.
If you wanted to do something easier to transition to, you would change each x in an x.x.x.x existing IPv4 address to 16 or 32 bits, so that every IPv4 address would _be_ a IP next generation address, so far as it appeared from a user or administrator perspective. Call it tuple based addressing - every address is internally a tuple of four integers (x,y,z,w), size of each component limited for convenience only.
Of course everyone would still have to upgrade all their routers and networking stacks to support a new header format before the plan would become fully effective. The difference is that existing addresses would be forward compatible _forever_. The numbering plan of the Internet would not need to be re-done from scratch. Instead of making everybody get two, incompatible "phone numbers" on two incompatible phone networks, everyone with an existing "phone number" could keep it indefinitely. That would make the transition go much smoother. Dual stack stinks - net admins have to maintain two independent numbering plans and there are always problems associated with choosing which stack to use to reach any given host. No way to know much of the time whether the other stack is functional or not.
Whereas once you were part of the next generation Internet with a scheme like this, you could reach _any_ host using a single, upgraded stack. Not just 198.55.55.55 (which could keep that address forever), but also 423.48.17.4 or 12.4.6.868.
The IPv6 transition will probably happen someday after immense pain and torment, but it could have been done years ago if the designers picked a transparent addressing plan compatible scheme instead of worrying about how many prefixes core routers could handle. With the scheme just outlined, the number of prefixes would be _identical_ to a comparable IPv4 network. The same prefixes we have now, just represented somewhat differently internally. Every core router on the planet would only need one routing database, not the one we have plus a new one in addition to that.
The routers don't deal in text format. They work with binary.
No kidding. However, millions of not billions of man hours have gone into configuring the current network addressing scheme throughout the world. That will all have to be redone, not to mention maintaining two different addressing schemes in parallel for the next decade or so.
Whatever its downsides, the suggested scheme would be nearly transparent upgrade on the system administration level. It could have all happened ten years ago and most people wouldn't even notice.
If this was done properly no renumbering would be required, the new addresses would be the same in text format as the old addresses. The only difference would be that each number in an IP address could go from 0..65535 instead of 0..255. The old address space would be a logical subset of a new address space, all existing hosts and subnets would have the same addresses (in text format), and so on.
That would be a *much* simpler transition than what we have planned now.
Patents are an orthogonal issue to whether or not the standard is open.
Add quotation marks there around "standard" and "open" and you have it just about right. As I said before the Internet has far higher standards for "openness" than some consortium of money grubbers like MPEG, ITU, or the ISO.
Or in other words, your preferred definition is slanted towards patent trolls, who like to pretend to the entire world that something they have government granted monopolies on is both "open" and "standard" merely because they came to an agreement with other patent trolls in a process fair and balanced to the interest of patent trolls and other leeches on the general welfare.
No one is forced to accept arbitrary coins of course, but that doesn't mean that coins produced from valuable materials do not have intrinsic value completely apart from the willingness of everyday vendors, merchants, and banks to accept it. If it is valuable, somebody will.
If gold didn't have value in jewelry, for example, it is much less likely that it would ever have become a monetary standard. Something isn't natural money unless it has intrinsic value apart from its use as money. Now if you have something rare and hard to produce, it might have value primarily as a monetary standard, but no one is going to start accepting it unless as you say someone mandates it as the "coin of the realm".
Where if you owe me $1,000,000 on the other hand I would be stupid not to accept payment in precious metals at market rates, as long as the market is liquid enough. The market for precious metals has always been far more liquid than just about anything else. Maybe not for purchasing candy at the corner store, but for large sums virtually anywhere and everywhere.
Column family based partitioning gives it a shade of column orientation I admit. That is a feature no relational database I know of really has, although it would be a great enhancement. The closest most relational databases come to column partitioning is out of line storage for BLOBs and CLOBs.
My complaint is that the "column orientation" of Cassandra is more like free form "field orientation". If row columns (fields) are optional, you are not really storing "columns" at all, but rather arbitrary sets of fields (including repeating groups) in rows or sub rows.
One hundred times earnings? Worse, actually. One hundred times revenue. They are probably barely making a profit at this point. Facebook isn't ready for an IPO at fifty billion dollars until it comes up with revenue on the order of two or three billion dollars a year, and profit of at least half that. Even then it might be a dodgy investment, but I have little doubt they could pull it off, IPO wise.
All money represents debt. By definition money represents a promise to pay by somebody else; if it didn't, it wouldn't be money.
It would be more accurate to say that paper or promissory currencies represent credit. The only debt is on the part of the currency issuer (including banks and any other issuer of promissory notes), not the currency holder.
Specie, on the other hand, is neither debt, nor credit, nor a promise to pay anyone or anybody, although it may have some usefulness in that regard. It is accepted rather because it has intrinsic value that no promissor can default on.
Welcome to the first five minutes of using a column store.
Calling Cassandra a "column store" or "column oriented database" is an abuse of the language. Real column oriented databases store "columns" of data in a linear sequential manner, so that they can be scanned in the fastest manner possible.
Cassandra isn't like that - it stores denormalized rows with repeating groups in a free form manner, not "columns" at all. If it were a real column oriented database it would be completely unusable for most online web and OLTP applications. Real column oriented databases are designed for OLAP and numerical-statistical analysis, getting a single "row" or "record", "record group", or "even" document at a time is the opposite of what they are designed for.
At best, you could call Cassandra a "cell" or "field" oriented database. But a "column oriented" database it is not. It is practically the opposite of a column oriented database.
H.264 is a theatrical production standard, a broadcast, a cable and sattelite distribution standard. H.264 is supported by every HDTV set, video game console and set top box on the planet.
That is a bit of an over-statement, but if you put quotation marks around "standard" you would have it just about right. Nothing patent encumbered can ever become a real standard, because the entire purpose of patents is to make sure that nothing ever does, not until all the pertinent patents expire at any rate.
I would think that readers of a site that roundly derided Microsoft's OOXML "standard" would understand the difference between an "open standard" and a "proprietary standard that's been dropped into the public domain but which has seen no significant attempts at standardization," but then, I guess when Google gets involved, all logic goes out the window.
Patents anyone? OOXML isn't ideal, but it is royalty free, which means for all its warts it is universally deployable. Any patent encumbered "standard" is not. The whole idea of a "patented standard" is a joke. OOXML is mom, baseball, and apple pie compared to the darkness that something like H.264 sheds over the landscape.
Furthermore the idea that something has to be developed by a "standards body" to be an "open standard" is to make a fetish out of international bureaucracy, one which is the enemy of the open Internet in this case. The Internet has far higher standards for "openness" than most putative "standards" bodies.
They certainly know better than all the H.264 shills out there. Patented "standards" like H.264 are the enemy. Patents are evil - the greatest impediment to the progress of science and the useful arts ever developed.
Google is simply saying, no we are not going to pay to distribute evil empire codecs like that. If you want to persuade every user on the planet to pony up for H.264 licensing fees, then more power to you. I would rather persuade the legislatures of the world to eliminate the plague of patents completely, or nearly so at any rate.
H.264 is an open standard - it went through the standardization process, with all the feedback solicitation and ratification that that implies
Any standards body that produces patent encumbered "standards" isn't to be taken seriously. Patents are the enemy of standardization, and so are "standards" bodies who endorse and promote them.
Until Google submits VP8 to ISO or some other standards body, it's not an "open" standard
Standards bodies (let alone evil empire standards bodies) do not have a monopoly on "openness". VP8 is open source, openly documented, and royalty free. That makes it orders of magnitude more "open" than H.264 is.
"Patented standard" is an oxymoron. The entire purpose of patents is to makes sure that nothing can ever become a real standard, but rather a private monopoly. To the degree that H.264 is a standard at all, it is the best example of an "evil empire" standard ever developed.
I don't see what is bad about handing off video/audio rendering to the OS frameworks
Perhaps because doing so in the Linux case quasi-legitimizes illegal behavior? I would be all for it, personally, if the preponderance of Linux users who want H.264 support would purchase a commercial plugin or otherwise pay the appropriate licensing fees. Since most apparently won't do that even if the option is provided to them, handing off H.264 support to be illegally implemented by the "operating system" is counterproductive in terms of promoting universally legally deployable standards.
So what? The GPL is nearly meaningless so far as third party patent encumbered software is concerned. No one can legally use x264 without paying patent royalties. Depending on MPEG-LA's licensing whims x264 might not be legal to distribute tomorrow in any case, at least in binary form.
Slashdot Mirror
What address does an IPv4 host use to respond to an IPv6 host after the IPv4 pool is exhausted? It can't be IPv4 because there are none for the IPv6 host to be assigned and it can't be IPv6 because IPv4 does not understand IPv6.
DJB is suggesting something like an IPv4 plus with a compatible, routable address space. Meaning those IPv4 hosts instead of being upgraded to dual stack would be upgraded to single stack IPv4 plus, no network reconfiguration or renumbering required.
Every IPv4 configuration in the world would be a viable IPv4 plus configuration. Major corporations could switch to IPv4 plus without knowing what happened. Network admins could be blissfully ignorant of the expanded address space for years.
After this silent deployment had taken place during the natural refresh cycle, IPv4 plus addresses in the extended address space would be reachable by virtually everybody and could be used in a _single stack_ configuration without major headaches.
As it happens there was such a proposal before the IETF back in 1993 (aka "TP/IX", RFC1475), but the IETF rejected it because they wanted a cleaner, "reboot the world" design. Bad mistake.
Any "fix" to this would require changing how IPv4 anyway, which is WORSE than switching to IPv6.
A proper fix to IPv4 would have the major advantage of being configuration compatible. A large corporation could upgrade to "IPv4 plus" and not know that it happened. It would just work. No dual stacking, no renumbering, no network reconfiguration, let alone maintaining two independent configurations at once.
Once most of the organizations in the world had IPv4 plus compatible devices (without even knowing about it) addresses in the extended area could be used. That would be a _much_ cleaner transition plan than the reboot the world plan associated with IPv6. It could have happened a decade ago and most people would neither know nor care.
There is a trivial mapping between IPv4 and IPv6 addresses
Yes, but it is not supposed to be one that anyone is supposed to use, except as a temporary crutch. The IETF wants the entire existing network configuration of the world to go away. That includes all v6 mapped v4 addresses, network prefixes, and so on.
A more viable transition plan would have the IPv6 backbone route those prefixes forever, as opposed to the current plan, which to first approximation is "not at all" - not on the v6 backbone at any rate.
The author claims that IPv6 should have been designed as an extension to IPv4 so that IPv4 and IPv6 hosts can communicate with each other directly. This is fundamentally impossible.
On the contrary, if IPv6 address space was designed to be an long term extension of the IPv4 address, dual stacking would not be required. Network reconfiguration wouldn't be required either.
DJB recognizes that of course everyone will eventually have to get upgrades to handle the extra address space, or for anyone to practically use an address in the extended area. The difference is that the upgrade would be transparent, because no network reconfiguration would have to be done. All old IP addresses and routing prefixes would be preserved, on the new network, forever.
After a few years, everyone would have extended address space compatible software and hardware and it would just start working automatically with extended addresses without any user intervention required. Network administrators, in particular, would be required to do precisely nothing. That is a major advantage in the real world.
position #1 being the recurring myth that ISPs are subject to common carrier regulations.
The only reason why ISPs are not subject to common carrier regulations is because the FCC is wildly (and opportunistically) misreading the Communications Act, which defines "Telecommunications" as follows:
Sounds like Internet access service, no? In fact there has _never_ been a better example of telecommunications service than Internet access service.
So while you are technically right as a matter of current federal regulation (which can be changed/corrected at any time by the FCC) that Internet access service is not subject to Title II common carrier regulations, all those Slashdot users you so blithely accuse are arguably right as a matter of law. It is only the worst kind of opportunistic FCC legerdemain that keeps the law from being enforced as written.
Title II is far more sound as a basis for net neutrality regulations than whatever crazy legal theory the FCC has been promoting lately to regulate broadband access while simultaneously pretending that it is not a form of "telecommunications". And if they are forced into that position, that is precisely what the FCC will do - read the law the way it was written, objections from all the people who are horrified at the prospect notwithstanding.
First of all, it is technically open source, but the license the community edition uses means it cannot legally be used by businesses.
That is not technically "open source" by any common definition, most particularly not that of the Open Source Initiative, which prohibits discrimination against fields of endeavor.
Please clarify what you mean by forever. The IPv4 to IPv6 mapping is forever. Nobody is going to come and steal away that little 32 bit slice of the v6 address space.
No, the addresses will simply be useless as soon as the IPv4 net either ceases to exist or ceases to be routed to the location of the addressee. Strictly a stop gap measure. The IPv4 numbering plan is going away, and that means that every public facing server on the planet will have to be dual stacked (with two completely independent addresses) for at least a decade.
You can reach any v4 host from v6 using this mapping using NAT64. And you will be able to do that as for as long anyone cares to keep the v4 net alive together with a NAT64 gateway.
Aye, that's the rub.
No, there is not actually any way v6 could have been constructed in a different way that would have made the ISPs of the world adopt it sooner. They will adopt when forced to, no sooner, no matter how it works.
Not true. The scheme I described, would allow existing networks to keep their IP addresses and network numbering plan forever, because the new numbering plan would be a transparent expansion of the old numbering plan. Every existing IP address, network, and subnet would be an IP next generation address, so far as the user could tell. No need renumber anything. No new DNS entries, no dual stack, etc. Over five or ten years, everyone would have IP next generation hardware and software and not even know it, or need to do anything to put it into effect. That includes ISPs and network administrators of all stripes.
An ISP could literally transition to IP next generation without knowing the transition took place, because they would not have to allocate replacement addresses for anyone. There wouldn't be two parallel networks, two parallel backbones, two sets of DNS entries for everything and everything.
Instead the natural upgrade cycle would have brought everyone everywhere into IP next generation compliance in less than a decade without them even knowing about it. And then as old style IP addresses were depleted, the few stragglers remaining would actually become aware they needed some sort of router upgrade. Not new IP addresses, not a new numbering plan, not new DNS entries, just a router upgrade. Done.
The entire transition would mostly be visible only to programmers and hardware engineers. Network administrators would hardly have to care. The transition, instead of being a world wide ten year headache, would pass with hardly anyone knowing about it. That is what a conservative addressing expansion plan would look like. No attempt to reboot the world from scratch.
They did this. Except they added 12 octets in front of v4 and mapped existing v4 addresses to 0.0.0.0.0.0.0.0.0.0.0.0.x.x.x.x.
A clean transition would have to support a conformal mapping of the current IPv4 address space forever. IPv6 isn't designed to do that, so the IPv4 mapping is useless.
If you wanted to do something easier to transition to, you would change each x in an x.x.x.x existing IPv4 address to 16 or 32 bits, so that every IPv4 address would _be_ a IP next generation address, so far as it appeared from a user or administrator perspective. Call it tuple based addressing - every address is internally a tuple of four integers (x,y,z,w), size of each component limited for convenience only.
Of course everyone would still have to upgrade all their routers and networking stacks to support a new header format before the plan would become fully effective. The difference is that existing addresses would be forward compatible _forever_. The numbering plan of the Internet would not need to be re-done from scratch. Instead of making everybody get two, incompatible "phone numbers" on two incompatible phone networks, everyone with an existing "phone number" could keep it indefinitely. That would make the transition go much smoother. Dual stack stinks - net admins have to maintain two independent numbering plans and there are always problems associated with choosing which stack to use to reach any given host. No way to know much of the time whether the other stack is functional or not.
Whereas once you were part of the next generation Internet with a scheme like this, you could reach _any_ host using a single, upgraded stack. Not just 198.55.55.55 (which could keep that address forever), but also 423.48.17.4 or 12.4.6.868.
The IPv6 transition will probably happen someday after immense pain and torment, but it could have been done years ago if the designers picked a transparent addressing plan compatible scheme instead of worrying about how many prefixes core routers could handle. With the scheme just outlined, the number of prefixes would be _identical_ to a comparable IPv4 network. The same prefixes we have now, just represented somewhat differently internally. Every core router on the planet would only need one routing database, not the one we have plus a new one in addition to that.
Have you ever read the tenants of the "church"...
FYI: churches do not have "tenants", they have "tenets". "Tenants" are for apartment buildings.
The routers don't deal in text format. They work with binary.
No kidding. However, millions of not billions of man hours have gone into configuring the current network addressing scheme throughout the world. That will all have to be redone, not to mention maintaining two different addressing schemes in parallel for the next decade or so.
Whatever its downsides, the suggested scheme would be nearly transparent upgrade on the system administration level. It could have all happened ten years ago and most people wouldn't even notice.
Qwest probably leases the lines and contracts the installations through AT&T
Qwest is the phone company here. They own the lines.
If this was done properly no renumbering would be required, the new addresses would be the same in text format as the old addresses. The only difference would be that each number in an IP address could go from 0..65535 instead of 0..255. The old address space would be a logical subset of a new address space, all existing hosts and subnets would have the same addresses (in text format), and so on.
That would be a *much* simpler transition than what we have planned now.
Patents are an orthogonal issue to whether or not the standard is open.
Add quotation marks there around "standard" and "open" and you have it just about right. As I said before the Internet has far higher standards for "openness" than some consortium of money grubbers like MPEG, ITU, or the ISO.
Or in other words, your preferred definition is slanted towards patent trolls, who like to pretend to the entire world that something they have government granted monopolies on is both "open" and "standard" merely because they came to an agreement with other patent trolls in a process fair and balanced to the interest of patent trolls and other leeches on the general welfare.
No one is forced to accept arbitrary coins of course, but that doesn't mean that coins produced from valuable materials do not have intrinsic value completely apart from the willingness of everyday vendors, merchants, and banks to accept it. If it is valuable, somebody will.
If gold didn't have value in jewelry, for example, it is much less likely that it would ever have become a monetary standard. Something isn't natural money unless it has intrinsic value apart from its use as money. Now if you have something rare and hard to produce, it might have value primarily as a monetary standard, but no one is going to start accepting it unless as you say someone mandates it as the "coin of the realm".
Where if you owe me $1,000,000 on the other hand I would be stupid not to accept payment in precious metals at market rates, as long as the market is liquid enough. The market for precious metals has always been far more liquid than just about anything else. Maybe not for purchasing candy at the corner store, but for large sums virtually anywhere and everywhere.
Column family based partitioning gives it a shade of column orientation I admit. That is a feature no relational database I know of really has, although it would be a great enhancement. The closest most relational databases come to column partitioning is out of line storage for BLOBs and CLOBs.
My complaint is that the "column orientation" of Cassandra is more like free form "field orientation". If row columns (fields) are optional, you are not really storing "columns" at all, but rather arbitrary sets of fields (including repeating groups) in rows or sub rows.
100x earnings... yeah, no bubble here.
One hundred times earnings? Worse, actually. One hundred times revenue. They are probably barely making a profit at this point. Facebook isn't ready for an IPO at fifty billion dollars until it comes up with revenue on the order of two or three billion dollars a year, and profit of at least half that. Even then it might be a dodgy investment, but I have little doubt they could pull it off, IPO wise.
All money represents debt. By definition money represents a promise to pay by somebody else; if it didn't, it wouldn't be money.
It would be more accurate to say that paper or promissory currencies represent credit. The only debt is on the part of the currency issuer (including banks and any other issuer of promissory notes), not the currency holder.
Specie, on the other hand, is neither debt, nor credit, nor a promise to pay anyone or anybody, although it may have some usefulness in that regard. It is accepted rather because it has intrinsic value that no promissor can default on.
Welcome to the first five minutes of using a column store.
Calling Cassandra a "column store" or "column oriented database" is an abuse of the language. Real column oriented databases store "columns" of data in a linear sequential manner, so that they can be scanned in the fastest manner possible.
Cassandra isn't like that - it stores denormalized rows with repeating groups in a free form manner, not "columns" at all. If it were a real column oriented database it would be completely unusable for most online web and OLTP applications. Real column oriented databases are designed for OLAP and numerical-statistical analysis, getting a single "row" or "record", "record group", or "even" document at a time is the opposite of what they are designed for.
At best, you could call Cassandra a "cell" or "field" oriented database. But a "column oriented" database it is not. It is practically the opposite of a column oriented database.
H.264 is a theatrical production standard, a broadcast, a cable and sattelite distribution standard. H.264 is supported by every HDTV set, video game console and set top box on the planet.
That is a bit of an over-statement, but if you put quotation marks around "standard" you would have it just about right. Nothing patent encumbered can ever become a real standard, because the entire purpose of patents is to make sure that nothing ever does, not until all the pertinent patents expire at any rate.
I would think that readers of a site that roundly derided Microsoft's OOXML "standard" would understand the difference between an "open standard" and a "proprietary standard that's been dropped into the public domain but which has seen no significant attempts at standardization," but then, I guess when Google gets involved, all logic goes out the window.
Patents anyone? OOXML isn't ideal, but it is royalty free, which means for all its warts it is universally deployable. Any patent encumbered "standard" is not. The whole idea of a "patented standard" is a joke. OOXML is mom, baseball, and apple pie compared to the darkness that something like H.264 sheds over the landscape.
Furthermore the idea that something has to be developed by a "standards body" to be an "open standard" is to make a fetish out of international bureaucracy, one which is the enemy of the open Internet in this case. The Internet has far higher standards for "openness" than most putative "standards" bodies.
We're Google, and we know better than all of you.
They certainly know better than all the H.264 shills out there. Patented "standards" like H.264 are the enemy. Patents are evil - the greatest impediment to the progress of science and the useful arts ever developed.
Google is simply saying, no we are not going to pay to distribute evil empire codecs like that. If you want to persuade every user on the planet to pony up for H.264 licensing fees, then more power to you. I would rather persuade the legislatures of the world to eliminate the plague of patents completely, or nearly so at any rate.
H.264 is an open standard - it went through the standardization process, with all the feedback solicitation and ratification that that implies
Any standards body that produces patent encumbered "standards" isn't to be taken seriously. Patents are the enemy of standardization, and so are "standards" bodies who endorse and promote them.
Until Google submits VP8 to ISO or some other standards body, it's not an "open" standard
Standards bodies (let alone evil empire standards bodies) do not have a monopoly on "openness". VP8 is open source, openly documented, and royalty free. That makes it orders of magnitude more "open" than H.264 is.
"Patented standard" is an oxymoron. The entire purpose of patents is to makes sure that nothing can ever become a real standard, but rather a private monopoly. To the degree that H.264 is a standard at all, it is the best example of an "evil empire" standard ever developed.
I don't see what is bad about handing off video/audio rendering to the OS frameworks
Perhaps because doing so in the Linux case quasi-legitimizes illegal behavior? I would be all for it, personally, if the preponderance of Linux users who want H.264 support would purchase a commercial plugin or otherwise pay the appropriate licensing fees. Since most apparently won't do that even if the option is provided to them, handing off H.264 support to be illegally implemented by the "operating system" is counterproductive in terms of promoting universally legally deployable standards.
So what? The GPL is nearly meaningless so far as third party patent encumbered software is concerned. No one can legally use x264 without paying patent royalties. Depending on MPEG-LA's licensing whims x264 might not be legal to distribute tomorrow in any case, at least in binary form.