Do you really think Amazon would risk reducing its selection ("Worlds best selection") and the massive amounts of money from potential buying customers to pursue competitions with other printers that will never bring in anywhere near the amount of money that the retail business does? Yeah, hmmm....
Because it's a circuit network and the provider controls its access points, so they can easily charge per connect (per call, per message). This is network non-neutrality but not on the Internet. On the internet, you get access and your price includes access for data rates and limits, what you do with that data and how (with your choice of apps) is up to you and cannot be taxed or surcharged. It's charging on perceived value, which rarely mirrors actual cost to the provider. If the Internet was like the cellular networks, *fill in your favorite snarky remark*.
Don't fool yourself; Google wants to control standards. Microsoft does it in a defacto and relatively closed way and crosses its fingers that it gets market buy in - Google sets up alliances of groups and leads them all to the watering hole. Microsoft's approach is definitely not open, and Google's is questionable (see OpenSocial). Open standards and free software are usually done through an open committee with many leading orgs, or through a grass roots open source effort with consensus or majority decision-making. Google's recent open efforts do not meet this criteria so far.
The idea is that all of your ec2 instances (virtual hosts) should be stateless, or lack any business state (they will have the state of the applications running on them though). The E in ec2 is elastic and the primary motivation behind being elastic is quickly scaling up AND down as your business needs change (minute by minute, hourly, daily, etc...) and keeping your business state on each of those instances wouldn't work for most cases.
So you need to keep it somewhere else, which is where S3 comes in, which is basically keeping your business state in the cloud of the internet, writable and readable from anywhere. It's true, relational and contentious data doesn't work well in S3, technically and the pricing model works against lots of little data with lots of reads-writes (its better for static and larger files). All this doesn't mean that you can't use s3 for some relational data though, a mysql driver was written for s3.
The fact is, you can still keep your business data outside of ec2, but it has the potential to now be the bottleneck and single point-of-failure for your online applications. You could even set up a set of ec2 instances to be your db hosts replicating between primary and secondaries, which probably wouldn't be any worse than most db replication solutions. Most db's can be designed to work around the bottleneck problem, and in tandem with s3 this might be quite a reasonable solution.
I say nay to the haterz:) ec2 allows you to put anything you want on your vhosts, so no persistent storage means its not an encouraged pattern.
The author of the article has really appropriated the field of computer science for his own eyebrow-raising purposes. It seems to be working with most respondents here though. He basically said CS is not convergent, its not a science, its not applied math, it involves a broader range of disciplines and will ultimately fade away. So the response from every sane person should be, "It is math you idiot. It is probability, it is linear algebra, it is discrete math!" And you can't argue that; this is computer science and you can't change that. As long as there are computers there will be a need for computer scientists.
His argument was really talking about software development in the face of industry, business and the almighty user. He should have been talking about how computer science departments are where most software developers and engineers are educated and that really isn't the right place for them. There should be a larger school of engineering and software development that teaches a broader set of topics for the ever broadening field of computation. This field will be technical, it will involve software and understanding of computer science, and the math that supports it. But it will also involve an understanding of other divergent fields that can be applied to the development of computer systems like information science, usability, anthropology, cognitive psychology, etc... There aren't too many reputable schools that do this well. Whenever I hear about this stuff, the graduates of these schools can't develop software worth a crap or design worth a crap. CS grads can't design worth a crap either because CS shouldn't teach that, it should be taught in software engineering and thats where the job demand is always going to be. There's always more engineers than scientists.
This article had no basis for its argument other than the opinion that Python programmers are smarter because they seek the language out (and the language rocks!). This whole argument belittles the fact that there are some amazingly talented Java programmers out there. There are great things going on in the world of Jakarta.apache.org and things like Hibernate and Spring rock! Hey aspect oriented programming was formalized by Java people.
Java is heavy handed, and for projects that don't require a week or more of design and analysis prep, Java is usually a dumb idea. It is weak in the thick client area, as well. But hey, Java is cool with me.
Let the language meet the need, and if you think something is cumbersome, I bet youre using it for the wrong purpose.
I was just in Guatemala, and although it depended on the city, the range of Internet costs at cafes was between 40 cents and 1$ US per hour. Many of these cafes were packed with Guatemalan youth and some adults, so its my thinking that the cost is reasonable for many people in the city there. I had even talked to people who said that the costs in the 90s were too high for anyone to go to the cafes, but now theyre much more reasonable.
I guess Nicaragua might be a different story, but those costs seem high.
Slashdot Mirror
Do you really think Amazon would risk reducing its selection ("Worlds best selection") and the massive amounts of money from potential buying customers to pursue competitions with other printers that will never bring in anywhere near the amount of money that the retail business does? Yeah, hmmm....
Because it's a circuit network and the provider controls its access points, so they can easily charge per connect (per call, per message). This is network non-neutrality but not on the Internet. On the internet, you get access and your price includes access for data rates and limits, what you do with that data and how (with your choice of apps) is up to you and cannot be taxed or surcharged. It's charging on perceived value, which rarely mirrors actual cost to the provider. If the Internet was like the cellular networks, *fill in your favorite snarky remark*.
Don't fool yourself; Google wants to control standards. Microsoft does it in a defacto and relatively closed way and crosses its fingers that it gets market buy in - Google sets up alliances of groups and leads them all to the watering hole. Microsoft's approach is definitely not open, and Google's is questionable (see OpenSocial). Open standards and free software are usually done through an open committee with many leading orgs, or through a grass roots open source effort with consensus or majority decision-making. Google's recent open efforts do not meet this criteria so far.
The idea is that all of your ec2 instances (virtual hosts) should be stateless, or lack any business state (they will have the state of the applications running on them though). The E in ec2 is elastic and the primary motivation behind being elastic is quickly scaling up AND down as your business needs change (minute by minute, hourly, daily, etc...) and keeping your business state on each of those instances wouldn't work for most cases.
:) ec2 allows you to put anything you want on your vhosts, so no persistent storage means its not an encouraged pattern.
So you need to keep it somewhere else, which is where S3 comes in, which is basically keeping your business state in the cloud of the internet, writable and readable from anywhere. It's true, relational and contentious data doesn't work well in S3, technically and the pricing model works against lots of little data with lots of reads-writes (its better for static and larger files). All this doesn't mean that you can't use s3 for some relational data though, a mysql driver was written for s3.
The fact is, you can still keep your business data outside of ec2, but it has the potential to now be the bottleneck and single point-of-failure for your online applications. You could even set up a set of ec2 instances to be your db hosts replicating between primary and secondaries, which probably wouldn't be any worse than most db replication solutions. Most db's can be designed to work around the bottleneck problem, and in tandem with s3 this might be quite a reasonable solution.
I say nay to the haterz
The author of the article has really appropriated the field of computer science for his own eyebrow-raising purposes. It seems to be working with most respondents here though. He basically said CS is not convergent, its not a science, its not applied math, it involves a broader range of disciplines and will ultimately fade away. So the response from every sane person should be, "It is math you idiot. It is probability, it is linear algebra, it is discrete math!" And you can't argue that; this is computer science and you can't change that. As long as there are computers there will be a need for computer scientists. His argument was really talking about software development in the face of industry, business and the almighty user. He should have been talking about how computer science departments are where most software developers and engineers are educated and that really isn't the right place for them. There should be a larger school of engineering and software development that teaches a broader set of topics for the ever broadening field of computation. This field will be technical, it will involve software and understanding of computer science, and the math that supports it. But it will also involve an understanding of other divergent fields that can be applied to the development of computer systems like information science, usability, anthropology, cognitive psychology, etc... There aren't too many reputable schools that do this well. Whenever I hear about this stuff, the graduates of these schools can't develop software worth a crap or design worth a crap. CS grads can't design worth a crap either because CS shouldn't teach that, it should be taught in software engineering and thats where the job demand is always going to be. There's always more engineers than scientists.
This article had no basis for its argument other than the opinion that Python programmers are smarter because they seek the language out (and the language rocks!). This whole argument belittles the fact that there are some amazingly talented Java programmers out there. There are great things going on in the world of Jakarta.apache.org and things like Hibernate and Spring rock! Hey aspect oriented programming was formalized by Java people. Java is heavy handed, and for projects that don't require a week or more of design and analysis prep, Java is usually a dumb idea. It is weak in the thick client area, as well. But hey, Java is cool with me. Let the language meet the need, and if you think something is cumbersome, I bet youre using it for the wrong purpose.
I was just in Guatemala, and although it depended on the city, the range of Internet costs at cafes was between 40 cents and 1$ US per hour. Many of these cafes were packed with Guatemalan youth and some adults, so its my thinking that the cost is reasonable for many people in the city there.
I had even talked to people who said that the costs in the 90s were too high for anyone to go to the cafes, but now theyre much more reasonable.
I guess Nicaragua might be a different story, but those costs seem high.