In our work ("Enterprise") setting, we have RAID-Z that does plenty of checksumming to survive failures; so we'd want the extra performance (the only reason where our work uses SSDs instead of spinning disks).
In a consumer setting; where people store precious irreplacable memories (pics of kids) and their ownly copy of financial data; it seems reliability should rule.
The biggest opportunity for GPUs in Databases isn't for "performance". As others pointed out - for performance it's easier to just throw money at the problem.
GPU powered databases do show promise for performance/Watt.
However, energy efficiency is not enough, energy proportionality is needed. The objective of this work is to create an entire platform that allows execution of GPU operators in an energy proportional DBMS, WattBD, and also a GPU Sort operator to prove that this new platform works. A different approach to integrate the GPU into the database has been used. Existing solutions to this problem aims to optimize specific areas of the DBMS, or provides extensions to the SQL language to specify GPU operation, thus, lacking flexibility to optimize all database operations, or provide transparency of the GPU execution to the user. This framework differs from existing strategies manipulating the creation and insertion of GPU operators directly into the query plan tree, allowing a more flexible and transparent framework to integrate new GPU-enabled operators. Results show that it was possible to easily develop a GPU sort operator with this framework. We believe that this framework will allow a new approach to integrate GPUs into existing databases, and therefore achieve more energy efficient DBMS.
Also note that you can write PostgreSQL stored procedures in OpenCL - which may be useful if you're doing something CPU intensive like storing images in a database and doing OCR or facial recognition on them:
http://wiki.postgresql.org/images/6/65/Pgopencl.pdf
Introducing PgOpenCL - A New PostgreSQL Procedural Language Unlocking the Power of the GPU
you so long as the information about you is perceived to have value.
The tricky part people are overlooking is value to whom?
Which basically means as long as you live,
Disagree - it basically means the information will be kept forever in certain silos. The NSA will probably keep your information forever - so 999 years ago of your descendants are suspected of something controversial then they can use what you post on/. to go after them. Health care & life insurance companies will probably keep some of your data forever so they set the rates for your descendants based on statistics from your genetic makeup. Google (or whomever buys them) will probably keep some of your information forever, just because they can.
But these will all be in private silos of data. The public internet will probably forget you much much much faster.
... poorly collected unreliable data also vanishes at at least the same rate (hopefully faster).
And assuming shoddy data disapears faster than good data, then the quality of available data should continually increase.
The majority of CPU cycles in data centers is going to be looking up and filtering specific records in database
Approximate Computing is especially interesting in databases.
One of the coolest projects in this space is Berkeley AMPLab's BlinkDB.
Their cannonical example
SELECT avg(sessionTime) FROM Table WHERE city='San Francisco' ERROR 0.1 CONFIDENCE 95%
should give you a good idea of how/why it's useful.
Their bencmarks show that Approximate Computing to 1% error is about 100X faster than Hive on Hadoop.
Now they'll know everything from when you wake up to when you take a bath.
If you turn the heater up, they'll probably start sending you more banner ads for cold medicines; and if you do it too often, they'll probably sell that data to your health & life insurance companies to raise your rates.
No thanks Google. Stop spying on us.
Isn't this almost exacty what Eminent Domain laws are designed for.
If some private company's blocking use of resources important to public or civic use (those cable right-of-ways) the government pretty much gets to take them and pay whatever it says they're worth.
Or do they only use those laws to kick out poor people for huge corporate developers?
Applies even more to internet hacking, where 4 of the 10 biggest hacker groups in the world are almost certaily DNI(CIA), DoD(DIA), DoJ(FBI), and DHS(NCSD). (probably all working under the alias "anonymous")
they can't fix a nearly ten year old bug with find.
You're welcome to submit the patch yourself if it bugs you that much.
Or you're welcome to pay someone to do it. Even them - and if you paid them enough, people'd complain less about the ratio of revenue they get from google.
And if the "bug" doesn't bug you enough to do either of the above, it's not really that important, is it?
I'd like to see a type system that can help enforce units (like mass * distance = force). If it were really lightweight (in typing) to create types with meaningful units, it could protect you from accidentally adding things like Mbits with MBytes.
because I've seen how much of a raging asshole Linuz can be.
So, fork it and run your own fork.
If enough people think like you (and think you're better than Linus), your fork will quickly reach a critical mass. Then you can either hire someone to deal with Linus, or ignore him altogether and let his followers seek out your patches to pull the part they want..
(and if you think I'm being sarcastic - I'm not - this is pretty much how a lot of the major distros work)
Opium production in Afghanistan has been on the rise since U.S. occupation started in 2001. Based on UNODC data, there has been more opium poppy cultivation in each of the past four growing seasons (2004â"2007) than in any one year during Taliban rule.
If its CIA - for doing their own drug smuggling, it's unlikely they'll blow their cover by sharing with your Oregon's PD.
If it's NYPD - they won't care outside of NY.
If it's NSA - they won't blow their cover for fear of more bad PR.
And that's just US agencies. Even if it is government, it's just as likely it's China's government. Or Singapore's. Or Russia's.
Or Afghanistan's, now that someone stopped the Taliban who were cracking down on Heroin.
And even if it is - wouldn't buyers and sellers take precautions to keep their privacy even from the guys (who are very likely criminals) running Silk Road anyway?
Simply check the 38 million for "password", "secret", and the username
If I have a password on file at Adobe (and I think I do), it's probably "password". And that's not a bad thing.
My email there is probably some variation of dontSpamMe@spam.la (that was such a wonderful service while it lasted); or some throwaway like adobespam@[my-own-domain].com.
And there's no way I ever gave them a credit card (or real name, etc).
Slashdot Mirror
enterprise ... consumer ... often benchmark better
Seems backwards.
In our work ("Enterprise") setting, we have RAID-Z that does plenty of checksumming to survive failures; so we'd want the extra performance (the only reason where our work uses SSDs instead of spinning disks).
In a consumer setting; where people store precious irreplacable memories (pics of kids) and their ownly copy of financial data; it seems reliability should rule.
performance ... put up cash...
The biggest opportunity for GPUs in Databases isn't for "performance". As others pointed out - for performance it's easier to just throw money at the problem.
GPU powered databases do show promise for performance/Watt.
http://hgpu.org/?p=8219
However, energy efficiency is not enough, energy proportionality is needed. The objective of this work is to create an entire platform that allows execution of GPU operators in an energy proportional DBMS, WattBD, and also a GPU Sort operator to prove that this new platform works. A different approach to integrate the GPU into the database has been used. Existing solutions to this problem aims to optimize specific areas of the DBMS, or provides extensions to the SQL language to specify GPU operation, thus, lacking flexibility to optimize all database operations, or provide transparency of the GPU execution to the user. This framework differs from existing strategies manipulating the creation and insertion of GPU operators directly into the query plan tree, allowing a more flexible and transparent framework to integrate new GPU-enabled operators. Results show that it was possible to easily develop a GPU sort operator with this framework. We believe that this framework will allow a new approach to integrate GPUs into existing databases, and therefore achieve more energy efficient DBMS.
Also note that you can write PostgreSQL stored procedures in OpenCL - which may be useful if you're doing something CPU intensive like storing images in a database and doing OCR or facial recognition on them: http://wiki.postgresql.org/images/6/65/Pgopencl.pdf
Introducing PgOpenCL - A New PostgreSQL Procedural Language Unlocking the Power of the GPU
you so long as the information about you is perceived to have value.
The tricky part people are overlooking is value to whom?
Which basically means as long as you live,
Disagree - it basically means the information will be kept forever in certain silos. The NSA will probably keep your information forever - so 999 years ago of your descendants are suspected of something controversial then they can use what you post on /. to go after them. Health care & life insurance companies will probably keep some of your data forever so they set the rates for your descendants based on statistics from your genetic makeup. Google (or whomever buys them) will probably keep some of your information forever, just because they can.
But these will all be in private silos of data. The public internet will probably forget you much much much faster.
Interesting and sad.
Those links would probably also be great for NSA & Google like data mining to monitize those users too.
... poorly collected unreliable data also vanishes at at least the same rate (hopefully faster). And assuming shoddy data disapears faster than good data, then the quality of available data should continually increase.
DB: Queries with Bounded Errors and Bounded Response Times on Very Large Data
The majority of CPU cycles in data centers is going to be looking up and filtering specific records in database
Approximate Computing is especially interesting in databases. One of the coolest projects in this space is Berkeley AMPLab's BlinkDB. Their cannonical example
should give you a good idea of how/why it's useful.
Their bencmarks show that Approximate Computing to 1% error is about 100X faster than Hive on Hadoop.
Now they'll know everything from when you wake up to when you take a bath. If you turn the heater up, they'll probably start sending you more banner ads for cold medicines; and if you do it too often, they'll probably sell that data to your health & life insurance companies to raise your rates. No thanks Google. Stop spying on us.
I'll bet Julian Assage .. Assage is
Why do so many forums (/., reddit) have so many Assage bashers even under articles that aren't about him.
Back when he was best known as being a postgres contributor, he seemed like a very normal nice guy.
It's not his fault that he claims big organizations are out to get him - because big organizations really are out to get him.
I live in Wisconsin, seriously, that "waste" heat is NOT wasted! It's freaking cold outside!! I'm an American, I want to be free to choose!
Even worse than that --- I have a number of friends who's rooftop solar produces more energy than they consume.
For them - the energy is "free" so nothing's wasted.
But instead they're forced to use the more environmentally harmful mercury-filled incandescants, or e-waste-with-dirty-manufacturing LED bulbs.
TL/DR: with rooftop solar, they banned the most environmentally friendly bulbs.
I was hoping the government would take the poles in their entirety, rather than trying to get them to share.
torches and pitchforks parade at the AT&T offices
Isn't this almost exacty what Eminent Domain laws are designed for. If some private company's blocking use of resources important to public or civic use (those cable right-of-ways) the government pretty much gets to take them and pay whatever it says they're worth. Or do they only use those laws to kick out poor people for huge corporate developers?
I'd rather see "click-to-pay" the default for javascript.
It's been many years since I've been annoyed by an irritating java applet, and there a few I find useful.
But ugh - so much javascript, and so many sites that practically require it.
Some kind of kinky "role-reversal play' among government agencies?
I imagine it's more some turf war / battle over budgets.
Remember back in in 2008, when the FBI wanted the right to monitor all internet traffic ("The surveillance should include all Internet traffic, Mueller said, whether it be .mil, .gov, .com--whichever network you're talking about.")? Apparently the NSA got an even bigger budget for that project than the FBI did, and I imagine the FBI's been jealous ever since.
Ever since news about how guys like Chalabi would play the State Department, Pentagon, and CIA off of each other, I've wondered how many of the world's conflicts are actually DNI(CIA) vs DoD(DIA)
Applies even more to internet hacking, where 4 of the 10 biggest hacker groups in the world are almost certaily DNI(CIA), DoD(DIA), DoJ(FBI), and DHS(NCSD). (probably all working under the alias "anonymous")
If it had an unlocked bootloader; and I could find a Debian / Mint / etc port for it; I probably would have bought one.
At least then if (when) I noticed RT sucking I would have known I could fall back to something I could use.
But they didn't. So I didn't.
they can't fix a nearly ten year old bug with find.
You're welcome to submit the patch yourself if it bugs you that much.
Or you're welcome to pay someone to do it. Even them - and if you paid them enough, people'd complain less about the ratio of revenue they get from google.
And if the "bug" doesn't bug you enough to do either of the above, it's not really that important, is it?
I'd like to see a type system that can help enforce units (like mass * distance = force). If it were really lightweight (in typing) to create types with meaningful units, it could protect you from accidentally adding things like Mbits with MBytes.
I worry about the fumes of a makerbot in the poorly ventilated classrooms in many schools.
At least if they put the 3D-printers in a shop class, they surely have better airflow.
because I've seen how much of a raging asshole Linuz can be.
So, fork it and run your own fork.
If enough people think like you (and think you're better than Linus), your fork will quickly reach a critical mass. Then you can either hire someone to deal with Linus, or ignore him altogether and let his followers seek out your patches to pull the part they want..
(and if you think I'm being sarcastic - I'm not - this is pretty much how a lot of the major distros work)
If its CIA - for doing their own drug smuggling, it's unlikely they'll blow their cover by sharing with your Oregon's PD.
If it's NYPD - they won't care outside of NY.
If it's NSA - they won't blow their cover for fear of more bad PR.
And that's just US agencies. Even if it is government, it's just as likely it's China's government. Or Singapore's. Or Russia's.
Or Afghanistan's, now that someone stopped the Taliban who were cracking down on Heroin.
And even if it is - wouldn't buyers and sellers take precautions to keep their privacy even from the guys (who are very likely criminals) running Silk Road anyway?
Simply check the 38 million for "password", "secret", and the username
If I have a password on file at Adobe (and I think I do), it's probably "password". And that's not a bad thing.
My email there is probably some variation of dontSpamMe@spam.la (that was such a wonderful service while it lasted); or some throwaway like adobespam@[my-own-domain].com.
And there's no way I ever gave them a credit card (or real name, etc).
TL/DR: password=password doesn't neccessarily imply insecure
If only it were open-sourced, then you (or your community) could pay third parties for support for as long as you wanted.
Most amazing amateur film I've ever seen: http://en.wikipedia.org/wiki/Star_Wreck:_In_the_Pirkinning And free (Creative Commons) licensed too.
More interesting than you think.
This might be exactly the right PR move that might make people appreciate NSA monitoring of the internet.
Hope their PR guys think about it.