That not true in general. It's only true for old fashioned "code forward" computing where your code is specifying what to do with the data. With connectionist approaches, genetic computing, etc (and natural evolution), it's often the data not the code that is in control, and techniques like this are usually used specifically because you don't know how to "intelligently" solve the problem, so you instead, in essence, feed the data into an architecture where it organizes itself.
That doesn't really happen. Yes, with neural networking or genetic operations, various potentially uncategorized data creates a "model," but the truth is that the model is little more than a statistical analysis of the data.
The data was chosen and extracted by people who understood what they wanted to analyze. The system did not initiate the operation or define the parameters.
Somehow, I think Google's compute clusters have more power than whatever AI lab you were looking at.
It isn't the "computing power" it is the nature of the computing system.
Also, it is not a machine simply learning how to think on its own - it relies on millions of people to sort, filter, and connect the data. Basically harvesting both data and semantic connections from humans.
We had this discussion a little while back. The mythical AI where machines "learn" how to "think" is a long way away or possibly impossible with current technology.
The appearance of intelligence is not intelligence. A recommendations system or search engine may appear intelligent, but the part of the system that processes information "intelligently" was programmed by a person who understood the process. The computer is merely following directions.
Some knowledge based algorithms seem unpredictable when given random data. This is not intelligence either, it is more a result of unintended consequence. You can go back and figure out why it acted a certain way.
I used to work at "AllTheWeb.com" and I tell you, basic good search is fairly pedestrian and "cuil" doesn't even have that. It is an interesting and exiting but utterly useless layout. When it does find "relevant" content, which it does a very poor job at BTW, it is hard to find it on the page.
Someone give me a few $mill in VC, and I'll make a better search engine.
So all those people I met in law school and still socialize aren't my friends? That's kind of depressing, maybe I should notify them about that.
I would bet that they are not. Invite them over, say you have a case of cold beer, and say "bring a paint brush, we're painting." A "friend" would absolutely be there. Will your lawyer friends? Cops? Huh! I doubt it.
One of my first jobs was a bank teller. Our passwords were sealed in an envelop, which we initialed, and locked in a vault which needed two keys to open.
If the two officers needed my password, they'd open the vault, open the envelope, breaking my seal (letting me off the hook of responsibility).
The administrators *need* access to the highest level of security. Maybe software and operating systems as a whole need to be rebuilt in the shape of a military complex where sensitive access does not have to be granted to the builders.
But, hey, even the builders see the vaults before they are used.
Psychopaths are people who use manipulation, violence and intimidation to control others and satisfy selfish needs. However, they have a chronic inability to feel guilt, remorse or anxiety about any of their actions.
Yes.
However, only a small percentage in these professions is a psychopath. Most professionals in these groups suffer from depression and guilt and remorse...which is why there is such high job dissatisfaction.
Know many cops? Know many lawyers? Have you ever seen cops "high five" for smashing a kids face into the ground? Have you ever seen lawyers gloat over threatening a person with a law suit just to shut him up and laugh about it?
I have, and if you knew how many sociopaths and psychopaths are in these fields you'd be scared shitless. Trust me, cops and lawyers are not your friends.
A spammer, most salesman, most cops, most all lawyers, in fact, anyone who's living involves taking advantage of another human being is, on some level, a psychopath.
Does this surprise me? No. Am I sad for the family, you bet.
We as a society celebrate monetary success above all. That is what psychopaths are best at.
Just remember the words of Chief Seattle: "We are all poor, because we are all honest."
If you learned anything from SCO, it should be that you should have a solid case before taking someone to court, or you're going to become familiar with such terms as "case dismissed with prejudice" and "paying court fees". Its unfortunate that there is no Linux support for this motherboard, but the guy should have returned the motherboard and purchased one from a vendor that is happy to support Linux.
You miss the point. The multi-national corporations have been attacking consumers and citizens in the courts and with laws and law suits for so long, it is time to fight back.
A class-action law suit would cost Foxconn money. A class action suit against Yahoo for DRM would cost them money. A class action suit against anyone who isn't on the up and up will cost them money.
If they are going to use the gristmill that is civil law against us, we have to fight back with the same tools. There are no "victories" to be had in civil court. only damage. In a class action suit against a big company individuals have a lot less to lose and a big company can only take so much damage. We have to be constant and consistent. Wars are no longer fought with guns, but with lawyers.
We should use the power of the courts to FORCE companies to support customer choice. Microsoft bribes them NOT to, we have to up cost of those decisions. Make no mistake, there is a lot at stake the next decade. Will we be in a corporate run 1984 or will we be able to break the fascist momentum?
If I have learned anything from SCO, it is that you don't even need evidence to start a law suit.
I say we get a bunch of that have been "harmed" by their false statements of ACPI support, and sue them for time and effort. We could get millions. 1/3 goes to lawyers, and 2/3 goes to open source projects.
We "reasonably" expected their claims to be truthful. When we asked them for a fix, they changed their claims. That's gotta be worth a lawyer's time!!!
You can't make a system that will not lose data, you can only make a system that knows the last save point of 100% integrity.
Even that is *much* harder than it looks if you've got to worry about the validity of the process that checks the integrity of a save point that's just been written. At some point, you've just got to hit and hope.
The PostgreSQL MVCC and the WAL stuff is an amazingly well done. It took a while to get there, but I'd call it one of the best implementation's I've seen in open source.
Can you list a few common examples where a "real" database would do something better that MySQL can not?
select * from nflx_movies where mid in (select movie from imdbmap) ;
This is a pathalogical problem for MySQL. Where as this clause executes very slowly on MySQL, Oracle, PostgreSQL, and other analyze that correctly and optimize it.
create index foo on mytable(bar);
In almost any database I can execute that without locking the table. Not on MySQL.
If you have a HUGE databse, you better not add an index while the system is running.
I would suggest you download or use PostgreSQL, THEN compare it to MySQL. Also, see if you can find a "LAN Times Guide to SQL' it is probably one of the best beginners books.
Oh, right. I didn't think I needed to because of it's install base, which I figure is an indicator that there's enough people out there that think it's good enough for their purposes:)
How many people use Windows? How many people voted for Bush? How many people believe in ID? A numbers argument is not valid.
I suppose abusing me personally is an easier thing than admitting MySQL might not suck, but don't worry - I don't take offense!
If you took my post as insulting, I apologize it wasn't meant to be.
I think blaming MySQL for lazy developers not bothering to learn the ins and outs of databases is a bit rich.
Normally I'd agree, but the rah rah rah crowd, insists that MySQL is good and often times, it is the only experience developers have with a SQL database.
the niche market of people that want a simple data in/data out system without requiring a fully blown database.
If you are in the position of using a database, there is no extra cost for using a "full blown" database, so why even bother using anything less?
Then nothing is wrong with using Mysql of OLTP loads on commodity hardware for massive scalability, because everyone who does that uses Innodb. Innodb ends up being faster than MyIsam for these loads due to the finer lock granularity and concurrent writes.
I would argue two things:
(1) There are better platforms for OLTP even depending on how you define it and your particular implementation.
(2) This is precisely the opposet point of this thread, that Drizzle, being even dumber database than MySQL makes no sense.
Dude, implementing some features slow down some operations
In some cases yes, in some cases no. I would not consider that an axiom.
Think about the relative speed of a DSP versus a general purpose CPU
That is, of course, an apples and oranges sort of thing. A DSP is not a simpler general purpose CPU, just as a general purpose CPU like the x86 line, is not a more complex dsp.
If you don't understand the problem, the solution is going to confuse you.
The ad hominem, the sure sign you are using a bogus argument.
. Its like the "what's this tool used for" segment of this old home where they trot out this weird looking tool and try to figure out what it could be used for. Only after being presented with the right problem does the strange looking monstrosity's purpose become clear.
I get a kick out of MySQL defenders, they use like "right tool for the job" arguments, but never quantify what makes MySQL the right tool.
Their argument is that "it can do it," never that it does a particular job best.
So, tell me, what does MySQL have to justify its use?
What you are saying is absolutely the way things are going. Modern web development frameworks are trying to move all of the data processing away from the data (e.g. Ruby On Rails), and then use the database as dumb storage.
If you use a database purely as dumb storage, MySQL is probably faster, but I think that if one considers overall performance of an application, it depends more on how you code around the I/O bottlenecks (e.g. stored procedures versus application servers or ORMs).
Personally, I have found that using an in-memory SQLite database instead of home-brewed data structures allows me to quickly code up complex data structures and then manipulate them with 4-line SQL statements instead of 1 page functions. Check out http://www.squidoo.com/sqlitehammer#module7016183 for an example.
SQLite is a cool program in its own way being an embedded SQL engine makes it useful.
It is sad that almost a whole generation of developers will have to relearn basic data theory simply because they don't make an effort to learn about real databases.
I'm saying they're creating a product to satisfy a market that (for better or worse) thinks it doesn't need any of those things.
What, exactly, don't they need and why? More importantly, what are the pros/cons of removing functionality?
My response, if you read it objectively, is "MySQL is good for some things, and this will be good for other things,
But you've offered no constructive argument to support your assertion.
It sort of sounds like you're mostly just raging against useless developers that don't want to learn SQL and are building their logic in code instead of the database, where possible. That's fair enough. But that's not really MySQL's fault
I beg to differ, in the last 10 years, MySQL's failings has led to developers, probably like yourself, thinking that SQL databases "get in the way" more than they help. I pin that on MySQL. Had MySQL been a better SQL database, developers would be more encouraged to exploit the capabilities of a "real" database. Instead, they've learned that MySQL sucks at doing things and end up doing it in the application, and figure all the other systems suck just as much.
However, I'm not sure you understand the goals of Drizzle, nor the problem that it addresses.
I am not sure how REMOVING functionality addresses problems in this case. It just seems stupid. mean, seriously, every server purchased in the last couple years has at least a couple gig of RAM. The few hundred K of executable code being removed is inconsequential.
Having the data itself split on different servers forces you to put some data aware logic in a middleware layer.
The aggregation logic, certainly, but the efficiency and flexibility of a real database is even more helpful in this case, not less.
You may not realize this but Drizzle is the the fruit of SunLabs employee. Sun. Makers of Solaris, Sparc, Java, Dtrace, and ZFS. They didn't skip any classes in Comp Sci. If you are going to make an intelligent criticism of a product, you have to first try and understand it, instead of making an emotional response to some stupid developers.
So you are saying that Sunlabs has never made stupid decisions?
Slashdot Mirror
That not true in general. It's only true for old fashioned "code forward" computing where your code is specifying what to do with the data. With connectionist approaches, genetic computing, etc (and natural evolution), it's often the data not the code that is in control, and techniques like this are usually used specifically because you don't know how to "intelligently" solve the problem, so you instead, in essence, feed the data into an architecture where it organizes itself.
That doesn't really happen. Yes, with neural networking or genetic operations, various potentially uncategorized data creates a "model," but the truth is that the model is little more than a statistical analysis of the data.
The data was chosen and extracted by people who understood what they wanted to analyze. The system did not initiate the operation or define the parameters.
Somehow, I think Google's compute clusters have more power than whatever AI lab you were looking at.
It isn't the "computing power" it is the nature of the computing system.
Also, it is not a machine simply learning how to think on its own - it relies on millions of people to sort, filter, and connect the data. Basically harvesting both data and semantic connections from humans.
That is not AI.
We had this discussion a little while back. The mythical AI where machines "learn" how to "think" is a long way away or possibly impossible with current technology.
The appearance of intelligence is not intelligence. A recommendations system or search engine may appear intelligent, but the part of the system that processes information "intelligently" was programmed by a person who understood the process. The computer is merely following directions.
Some knowledge based algorithms seem unpredictable when given random data. This is not intelligence either, it is more a result of unintended consequence. You can go back and figure out why it acted a certain way.
I used to work at "AllTheWeb.com" and I tell you, basic good search is fairly pedestrian and "cuil" doesn't even have that. It is an interesting and exiting but utterly useless layout. When it does find "relevant" content, which it does a very poor job at BTW, it is hard to find it on the page.
Someone give me a few $mill in VC, and I'll make a better search engine.
So all those people I met in law school and still socialize aren't my friends? That's kind of depressing, maybe I should notify them about that.
I would bet that they are not. Invite them over, say you have a case of cold beer, and say "bring a paint brush, we're painting." A "friend" would absolutely be there. Will your lawyer friends? Cops? Huh! I doubt it.
One of my first jobs was a bank teller. Our passwords were sealed in an envelop, which we initialed, and locked in a vault which needed two keys to open.
If the two officers needed my password, they'd open the vault, open the envelope, breaking my seal (letting me off the hook of responsibility).
IT has to learn from banks.
The administrators *need* access to the highest level of security. Maybe software and operating systems as a whole need to be rebuilt in the shape of a military complex where sensitive access does not have to be granted to the builders.
But, hey, even the builders see the vaults before they are used.
Psychopaths are people who use manipulation, violence and intimidation to control others and satisfy selfish needs. However, they have a chronic inability to feel guilt, remorse or anxiety about any of their actions.
Yes.
However, only a small percentage in these professions is a psychopath. Most professionals in these groups suffer from depression and guilt and remorse...which is why there is such high job dissatisfaction.
Know many cops? Know many lawyers? Have you ever seen cops "high five" for smashing a kids face into the ground? Have you ever seen lawyers gloat over threatening a person with a law suit just to shut him up and laugh about it?
I have, and if you knew how many sociopaths and psychopaths are in these fields you'd be scared shitless. Trust me, cops and lawyers are not your friends.
You are an idiot, Did yo even think about what you wrote at all?
Like that sentence will make someone take you seriously.
First, spamming doesn't take advantage of anybody,
Seriously? Who pays for the wasted bandwidth of spam? Do you really think those adds are for reputable products?
Rant all you want. While some hyperbole is involved with most all slashdot posts, not so much as you protest, I think in this one.
A spammer, most salesman, most cops, most all lawyers, in fact, anyone who's living involves taking advantage of another human being is, on some level, a psychopath.
Does this surprise me? No. Am I sad for the family, you bet.
We as a society celebrate monetary success above all. That is what psychopaths are best at.
Just remember the words of Chief Seattle: "We are all poor, because we are all honest."
If you learned anything from SCO, it should be that you should have a solid case before taking someone to court, or you're going to become familiar with such terms as "case dismissed with prejudice" and "paying court fees". Its unfortunate that there is no Linux support for this motherboard, but the guy should have returned the motherboard and purchased one from a vendor that is happy to support Linux.
You miss the point. The multi-national corporations have been attacking consumers and citizens in the courts and with laws and law suits for so long, it is time to fight back.
A class-action law suit would cost Foxconn money. A class action suit against Yahoo for DRM would cost them money. A class action suit against anyone who isn't on the up and up will cost them money.
If they are going to use the gristmill that is civil law against us, we have to fight back with the same tools. There are no "victories" to be had in civil court. only damage. In a class action suit against a big company individuals have a lot less to lose and a big company can only take so much damage. We have to be constant and consistent. Wars are no longer fought with guns, but with lawyers.
We should use the power of the courts to FORCE companies to support customer choice. Microsoft bribes them NOT to, we have to up cost of those decisions. Make no mistake, there is a lot at stake the next decade. Will we be in a corporate run 1984 or will we be able to break the fascist momentum?
If I have learned anything from SCO, it is that you don't even need evidence to start a law suit.
I say we get a bunch of that have been "harmed" by their false statements of ACPI support, and sue them for time and effort. We could get millions. 1/3 goes to lawyers, and 2/3 goes to open source projects.
We "reasonably" expected their claims to be truthful. When we asked them for a fix, they changed their claims. That's gotta be worth a lawyer's time!!!
You can't make a system that will not lose data, you can only make a system that knows the last save point of 100% integrity.
Even that is *much* harder than it looks if you've got to worry about the validity of the process that checks the integrity of a save point that's just been written. At some point, you've just got to hit and hope.
The PostgreSQL MVCC and the WAL stuff is an amazingly well done. It took a while to get there, but I'd call it one of the best implementation's I've seen in open source.
Soon Davidson will be able to spam martians.
Can you list a few common examples where a "real" database would do something better that MySQL can not?
select * from nflx_movies where mid in (select movie from imdbmap) ;
This is a pathalogical problem for MySQL. Where as this clause executes very slowly on MySQL, Oracle, PostgreSQL, and other analyze that correctly and optimize it.
create index foo on mytable(bar);
In almost any database I can execute that without locking the table. Not on MySQL.
If you have a HUGE databse, you better not add an index while the system is running.
I would suggest you download or use PostgreSQL, THEN compare it to MySQL. Also, see if you can find a "LAN Times Guide to SQL' it is probably one of the best beginners books.
Oh, right. I didn't think I needed to because of it's install base, which I figure is an indicator that there's enough people out there that think it's good enough for their purposes :)
How many people use Windows? How many people voted for Bush? How many people believe in ID? A numbers argument is not valid.
I suppose abusing me personally is an easier thing than admitting MySQL might not suck, but don't worry - I don't take offense!
If you took my post as insulting, I apologize it wasn't meant to be.
I think blaming MySQL for lazy developers not bothering to learn the ins and outs of databases is a bit rich.
Normally I'd agree, but the rah rah rah crowd, insists that MySQL is good and often times, it is the only experience developers have with a SQL database.
the niche market of people that want a simple data in/data out system without requiring a fully blown database.
If you are in the position of using a database, there is no extra cost for using a "full blown" database, so why even bother using anything less?
So while our web developers use mySQl, theirs use PostgreSQL!
Then nothing is wrong with using Mysql of OLTP loads on commodity hardware for massive scalability, because everyone who does that uses Innodb. Innodb ends up being faster than MyIsam for these loads due to the finer lock granularity and concurrent writes.
I would argue two things:
(1) There are better platforms for OLTP even depending on how you define it and your particular implementation.
(2) This is precisely the opposet point of this thread, that Drizzle, being even dumber database than MySQL makes no sense.
What is wrong with the data integrity of Innodb?
Nothing if you can deal with the performance hit.
Massively scalable OLTP loads on commodity hardware.
Well, "everything" is commodity hardware these days.
OLTP? MySQL? Not if I need to depend on data integrity, no thanks.
If he's smart, he's saved some real money in some hidden bank accounts and will exit the country through Mexico's revolving door in Tijuana.
A few million bucks, converted to euros, will buy a pretty good "rest of your life" in central and south america.
Also, did anyone notice a spike in spam this morning?
Dude, implementing some features slow down some operations
In some cases yes, in some cases no. I would not consider that an axiom.
Think about the relative speed of a DSP versus a general purpose CPU
That is, of course, an apples and oranges sort of thing. A DSP is not a simpler general purpose CPU, just as a general purpose CPU like the x86 line, is not a more complex dsp.
If you don't understand the problem, the solution is going to confuse you.
The ad hominem, the sure sign you are using a bogus argument.
. Its like the "what's this tool used for" segment of this old home where they trot out this weird looking tool and try to figure out what it could be used for. Only after being presented with the right problem does the strange looking monstrosity's purpose become clear.
I get a kick out of MySQL defenders, they use like "right tool for the job" arguments, but never quantify what makes MySQL the right tool.
Their argument is that "it can do it," never that it does a particular job best.
So, tell me, what does MySQL have to justify its use?
What you are saying is absolutely the way things are going. Modern web development frameworks are trying to move all of the data processing away from the data (e.g. Ruby On Rails), and then use the database as dumb storage.
If you use a database purely as dumb storage, MySQL is probably faster, but I think that if one considers overall performance of an application, it depends more on how you code around the I/O bottlenecks (e.g. stored procedures versus application servers or ORMs).
Personally, I have found that using an in-memory SQLite database instead of home-brewed data structures allows me to quickly code up complex data structures and then manipulate them with 4-line SQL statements instead of 1 page functions. Check out http://www.squidoo.com/sqlitehammer#module7016183 for an example.
SQLite is a cool program in its own way being an embedded SQL engine makes it useful.
It is sad that almost a whole generation of developers will have to relearn basic data theory simply because they don't make an effort to learn about real databases.
I'm saying they're creating a product to satisfy a market that (for better or worse) thinks it doesn't need any of those things.
What, exactly, don't they need and why? More importantly, what are the pros/cons of removing functionality?
My response, if you read it objectively, is "MySQL is good for some things, and this will be good for other things,
But you've offered no constructive argument to support your assertion.
It sort of sounds like you're mostly just raging against useless developers that don't want to learn SQL and are building their logic in code instead of the database, where possible. That's fair enough. But that's not really MySQL's fault
I beg to differ, in the last 10 years, MySQL's failings has led to developers, probably like yourself, thinking that SQL databases "get in the way" more than they help. I pin that on MySQL. Had MySQL been a better SQL database, developers would be more encouraged to exploit the capabilities of a "real" database. Instead, they've learned that MySQL sucks at doing things and end up doing it in the application, and figure all the other systems suck just as much.
However, I'm not sure you understand the goals of Drizzle, nor the problem that it addresses.
I am not sure how REMOVING functionality addresses problems in this case. It just seems stupid. mean, seriously, every server purchased in the last couple years has at least a couple gig of RAM. The few hundred K of executable code being removed is inconsequential.
Having the data itself split on different servers forces you to put some data aware logic in a middleware layer.
The aggregation logic, certainly, but the efficiency and flexibility of a real database is even more helpful in this case, not less.
You may not realize this but Drizzle is the the fruit of SunLabs employee. Sun. Makers of Solaris, Sparc, Java, Dtrace, and ZFS. They didn't skip any classes in Comp Sci. If you are going to make an intelligent criticism of a product, you have to first try and understand it, instead of making an emotional response to some stupid developers.
So you are saying that Sunlabs has never made stupid decisions?