Mac could go with only two buttons ("play" and "order a new battery from apple"), but Microsoft is stuck with at least four ("play", "reboot", "reinstall" and "upgrade").
Only Amazon.com could possibly come with a single button operation... but wait, don't they already have a patent on this?
Windows has been greatly improved over the years, but AS/400 has a much more impressive track record. If you can afford a small cluster with AS/400, and if you are comfortable with this technology, I think this would be a very good scenario. Let's not forget all the headaches that come with Windows (virus, patch, upgrade here and then...)
I would advise you to keep room in your budget for technical assistance in the configuration of your cluster, as this is a very critical system. Get some help from IBM for the setup, but right from the start explain to them that you want to maintain the system yourself and that you are not interested in outsourcing it, because for sure they will try to get you to sign a long-term contract.
You'll need to be very strong about this, as their reps are very good with scaring management to get a contract. Your best strategy here is to include, from the start, a budget for IBM training (which is excellent); this way, they will probably focus their commercial tactics on getting you in the best training available instead of trying to coerce you into outsourcing. In any case, whatever effort they do to convince your boss that maintenance require expertise will probably help you to get a better training.
Also don't get cheap on the extended warranty, get the best one. The odds are that the on-site service from IBM will come handy at some point. In the long run you'll sleep better.
So the conclusion would be: buying a cluster is a good choice, whatever the OS, but don't throw all your money in hardware, keep some for the setup assistance, for training and for a good extended warranty. Include those items in the servers budget, don't try to shove it in another cost center.
No one said it is more "secure", but it surely is more fault-tolerant. A cluster will offer uninterrupted service as long as there is one of the nodes that is still live, just like a RAID mirror. So if you have a small cluster of three servers and one machine goes down, you have plenty of time to either restart or replace it, with no downtime (but less performance of course).
If you have a single AS/400, the second it goes down the service is interrupted.
So the more servers you have in your pool, the more fault-tolerant you are (as long as the servers in your pool are decent, of course). If your budget for the hardware+OS is 100k, you would get a more fault-tolerant solution with a cluster of 5 20k servers than with a 100k monster.
Also a cluster is more scalable, because you can either add servers or upgrade the existing servers with no downtime. It's easy to pull out a machine, add a new hard drive or some memory, and push it back; nobody will notice (unless all your CPUs are constanly running at 90%). But if you want to upgrade your AS/400, first of all you need a business case, and then you need a good banker:-)
without knowing more of the specifics I'd probably recommend the Windows boxen - or, more likely, a cluster of them
I completely agree. I would use a Windows cluster instead of a single iSeries. But then in the mid-range class the IBM servers are shitty, so I'd go with HP. This setup would be far less expensive and much easier to upgrade so you don't face the same migration issue in a couple of years.
This isn't true. For mid-sized business, Great Plains is a good product, not as expensive as other ERP, with a solid customer base. And mid-sized business is a serious market; the economy is not driven by the Fortune 500, pal.
...rock-solid reliability, high-volume throughput and widespread realtime access to the data and application...
Wow, you sound just like a sales rep trying to buzzword his way to a big commission. Or like a IT project lead that has no budget issue and that is trying to sell the latest gizmo to the V-P.
You get a good, quiet PSU with the case, and there is even a knob where you can set the fan speed. The hd are mounted on rubber washers, not directly on metal, so even at full spin they are not noisy. You also get a heat duct that drains the cpu heat outside the case.
The Sonata 2 ships with one 120mm fan, I advise you to add a second (plenty of sockets on the case). Big fans turn slowly, which is less noisy.
This is what I have at home, and if you unplug the blue leds there is no way to tell if the power is on unless you put your hand behind the psu fan.
This being said, I advise you to always put the psu at the top of your list when you buy a pc. Good, reliable power will give your hd a longer life.
If you see Delphi and Access as "new" technologies, then yes, maybe it is time you retire. Unless your employer needs your assembly skills to debug the cofee machine.
(for your convenience I posted this as "plain old text")
If you don't have Canadian and/or American experience and credentials, most employers will just delete your resume
Suppose that you are a Canadian employer and you have to hire a new systems administrator. If you want to hire a foreign worker, would you choose someone with a PHD from the University of Ouagadougou (http://www.univ-ouaga.bf/) or someone with a bachelor degree from the Wichita State University (http://www.wichita.edu?
There are two challenges with the candidate from Burkina Faso. First the employer must find a way to confirm that he studied at the Ouagadougou University. Then he has to find reliable metrics to validate the quality of the education.
On the other hand, there is none of these issues with the candidate from WSU.
So if the WSU candidate is ok, why bother? What added value would an employer get from the other candidate? Cultural insight?
Go with unattended setups. You can add the drivers for all the hardware available, and you can plugin MSI to install additional software. If you have some software that has no MSI setup, you can build your own from a fresh install (see Microsoft documentation).
Plenty of documentation is available on the Microsoft website and there is no extra licensing.
SQL that is portable between PostgreSQL, DB2 and Oracle, and easily adaptible to Transact-SQL...
Easily adaptable? No it is not. You don't have table heritance or array fields in Oracle, which you have in Postgres; how do you "port" the queries? And T-SQL is quite different from the Oracle or Postgres SQL, even a junior programmer can tell that queries on one database are not easily adaptable to another. Which is why I suspect that you don't have much experience.
I gave a simple and concrete example of something that takes two minutes to implement in SQL and 2 days to implement using EJBs
No you did not. Your evaluation of how long this kind of implementation takes is wrong, so all your logic (which is based on that evaluation) is flawed. A simple EJB can be done in less than two minutes with JBuilder or any other J2EE IDE. Most of the code is generated by the software, not by the programmer.
An EJB is much easier to maintain than a SQL query because with EJBs you don't have to rewrite at all if you change the back-end, as long as the business logic is the same.
Referring to a database as "data persistence" shows ignorance of at least the reporting capabilities of a real database, and likely other things as well
According to Merriam-Webster, a database is "a usually large collection of data organized especially for rapid search and retrieval (as by a computer)". I don't see reporting capabilities, and I don't see why "data persistence" is not a valid description.
I suspect that when you say "real database" you apply a selective filter over the pool of existing products, and that you keep only the products having peripheral software. Which does not make the other products less of a database. Also, if reporting capabilities were included in a database, why would Oracle license Oracle Reports separately?
This being said, I find it obvious that either you have not been involved in a major software development project, or that you have been involved in poorly managed projects only. Applying a layer of abstraction between the application and the database is a very good practice and it saves lots of work whenever the application or the database has to be modified.
What do you do if someday for a commercial purpose you must change your database vendor? With your strategy you have to rewrite loads and loads of code, as each vendor has specific SQL features. But if you have an abstraction layer (like EJB) you basically change the drivers and, maybe, update the XML schema.
then next time we try to get rid of this relationship with your ridiculous country, don't come in Montreal a few days before the referendum for a "love-in".
and try to convince your corrupt government not to speed up the processing of immigration in the months before the referendum just to get a few thousand more votes for this Good Ol' Canada like they did in 95.
and while you are at it, pay yourself for the visits of your queen of england which we don't give a damn about, and give us back our share of the money wasted on your 2 Billions $ firearm program which is a big joke.
A lot of techies think that SQL skills are necessary to get the best out of an application. This might be true on small projects with little coding required, but for big systems it is safer to rely on a good data model, a good XML mapping and a good EJB server that will make an optimal use of such mapping. See Amazon.com, which is a textbook example of J2EE technologies performance on large-scale systems.
You'd rather code yourself all the interactions with the storage media instead of trusting a technology that has been specifically designed to understand a data structure and make the best use of it? Then how can you trust a RDBMS to handle the interaction with the filesystem? And then, how could you trust the o/s to deal with the hardware? At some point you have to rely on the software.
I know, it's difficult to let go those lovely queries. But soon SQL will be deprecated, just like COBOL, RPG, Clipper, Fortran, QBasic and others. Either you take the new wave, or you become a "legacy expert" and wish hard that there will be another Y2K-ish scam someday.
One just can't buy an Oracle database license. Accountants, lawyers and engineers must be involved to find out what insane amounts you will have to feed the monster every year. Basically, it's cheaper to have a SQL Server 4-processor license than hiring the experts required to calculate your Oracle yearly premium for the same hardware.
So my guess is that either the CFOs feel bad about selling incredibly overpriced products, or they just plain don't understand how the hell they can manage all these crazy contracts.
When Microsoft licensing is the low-cost alternative to your product, there is something terribly wrong.
Anyone who's authored a proper application with a database behind it SHOULD be thinking in SQL
This is a wrong statement. Data persistence does not have to be bound to the storage technology. Please take time to read about a technology before making so bold statements.
In the world of J2EE you can use container-managed persistence on entity beans (a flavor of EJB). Such object will handle the interaction with the database for you so you don't have to worry about SQL. And it will also address transaction issues so no need to fiddle with manual rollbacks.
With CMP, instead of fooling around with SQL queries you deal with business objects. Why bother with "update inventory" or "insert into cashRegister" when you can call methods on objects, like inventoryItem.substract() or cashRegister.feed() ? This provides you with a good layer between the application and the database, so whenever one has to change you don't have to mess with both. All you got to do to have all this magic is to keep some XML configuration up to date, and if you have the big bucks you can even get JBuilder to do it for you!
SQL is sooo yesterday. Stop wasting time on technical issues: add business value to your applications instead. And with JBoss you don't even have to pay for a good EJB container. (Of course you'll need a good CPU and a little more RAM, but hey, magic has a price!).
Take this statement: "All rules have an exception".
If this statement is true, then this rule itself must have an exception, in which case the statement is not true. However, if the rule itself is the only one without an exception, then the statement is making itself true by not having an exception.
What is a paradox? A simple demonstration that human logic is flawed.
Slashdot Mirror
Microsoft can't win this war of the buttons.
Mac could go with only two buttons ("play" and "order a new battery from apple"), but Microsoft is stuck with at least four ("play", "reboot", "reinstall" and "upgrade").
Only Amazon.com could possibly come with a single button operation... but wait, don't they already have a patent on this?
Yeah, there is no way to "exploit" anything on those versions...
Windows has been greatly improved over the years, but AS/400 has a much more impressive track record. If you can afford a small cluster with AS/400, and if you are comfortable with this technology, I think this would be a very good scenario. Let's not forget all the headaches that come with Windows (virus, patch, upgrade here and then...)
I would advise you to keep room in your budget for technical assistance in the configuration of your cluster, as this is a very critical system. Get some help from IBM for the setup, but right from the start explain to them that you want to maintain the system yourself and that you are not interested in outsourcing it, because for sure they will try to get you to sign a long-term contract.
You'll need to be very strong about this, as their reps are very good with scaring management to get a contract. Your best strategy here is to include, from the start, a budget for IBM training (which is excellent); this way, they will probably focus their commercial tactics on getting you in the best training available instead of trying to coerce you into outsourcing. In any case, whatever effort they do to convince your boss that maintenance require expertise will probably help you to get a better training.
Also don't get cheap on the extended warranty, get the best one. The odds are that the on-site service from IBM will come handy at some point. In the long run you'll sleep better.
So the conclusion would be: buying a cluster is a good choice, whatever the OS, but don't throw all your money in hardware, keep some for the setup assistance, for training and for a good extended warranty. Include those items in the servers budget, don't try to shove it in another cost center.
No one said it is more "secure", but it surely is more fault-tolerant. A cluster will offer uninterrupted service as long as there is one of the nodes that is still live, just like a RAID mirror. So if you have a small cluster of three servers and one machine goes down, you have plenty of time to either restart or replace it, with no downtime (but less performance of course).
:-)
If you have a single AS/400, the second it goes down the service is interrupted.
So the more servers you have in your pool, the more fault-tolerant you are (as long as the servers in your pool are decent, of course). If your budget for the hardware+OS is 100k, you would get a more fault-tolerant solution with a cluster of 5 20k servers than with a 100k monster.
Also a cluster is more scalable, because you can either add servers or upgrade the existing servers with no downtime. It's easy to pull out a machine, add a new hard drive or some memory, and push it back; nobody will notice (unless all your CPUs are constanly running at 90%). But if you want to upgrade your AS/400, first of all you need a business case, and then you need a good banker
I completely agree. I would use a Windows cluster instead of a single iSeries. But then in the mid-range class the IBM servers are shitty, so I'd go with HP. This setup would be far less expensive and much easier to upgrade so you don't face the same migration issue in a couple of years.
This isn't true. For mid-sized business, Great Plains is a good product, not as expensive as other ERP, with a solid customer base. And mid-sized business is a serious market; the economy is not driven by the Fortune 500, pal.
Wow, you sound just like a sales rep trying to buzzword his way to a big commission. Or like a IT project lead that has no budget issue and that is trying to sell the latest gizmo to the V-P.
You get a good, quiet PSU with the case, and there is even a knob where you can set the fan speed. The hd are mounted on rubber washers, not directly on metal, so even at full spin they are not noisy. You also get a heat duct that drains the cpu heat outside the case.
The Sonata 2 ships with one 120mm fan, I advise you to add a second (plenty of sockets on the case). Big fans turn slowly, which is less noisy.
This is what I have at home, and if you unplug the blue leds there is no way to tell if the power is on unless you put your hand behind the psu fan.
This being said, I advise you to always put the psu at the top of your list when you buy a pc. Good, reliable power will give your hd a longer life.
U=RI
I LOVE Kevin Costner. He was great in "The Dentist" and "LA Law".
Wait a minute... that was Corbin Bernsen. So I guess you are right, nobody likes Kevin Costner.
No, they just have to wait until Feb. 6, 2040
If you see Delphi and Access as "new" technologies, then yes, maybe it is time you retire. Unless your employer needs your assembly skills to debug the cofee machine.
(for your convenience I posted this as "plain old text")
Suppose that you are a Canadian employer and you have to hire a new systems administrator. If you want to hire a foreign worker, would you choose someone with a PHD from the University of Ouagadougou (http://www.univ-ouaga.bf/) or someone with a bachelor degree from the Wichita State University (http://www.wichita.edu?
There are two challenges with the candidate from Burkina Faso. First the employer must find a way to confirm that he studied at the Ouagadougou University. Then he has to find reliable metrics to validate the quality of the education.
On the other hand, there is none of these issues with the candidate from WSU.
So if the WSU candidate is ok, why bother? What added value would an employer get from the other candidate? Cultural insight?
Go with unattended setups. You can add the drivers for all the hardware available, and you can plugin MSI to install additional software. If you have some software that has no MSI setup, you can build your own from a fresh install (see Microsoft documentation).
Plenty of documentation is available on the Microsoft website and there is no extra licensing.
Easily adaptable? No it is not. You don't have table heritance or array fields in Oracle, which you have in Postgres; how do you "port" the queries? And T-SQL is quite different from the Oracle or Postgres SQL, even a junior programmer can tell that queries on one database are not easily adaptable to another. Which is why I suspect that you don't have much experience.
I gave a simple and concrete example of something that takes two minutes to implement in SQL and 2 days to implement using EJBs
No you did not. Your evaluation of how long this kind of implementation takes is wrong, so all your logic (which is based on that evaluation) is flawed. A simple EJB can be done in less than two minutes with JBuilder or any other J2EE IDE. Most of the code is generated by the software, not by the programmer.
An EJB is much easier to maintain than a SQL query because with EJBs you don't have to rewrite at all if you change the back-end, as long as the business logic is the same.
According to Merriam-Webster, a database is "a usually large collection of data organized especially for rapid search and retrieval (as by a computer)". I don't see reporting capabilities, and I don't see why "data persistence" is not a valid description.
I suspect that when you say "real database" you apply a selective filter over the pool of existing products, and that you keep only the products having peripheral software. Which does not make the other products less of a database. Also, if reporting capabilities were included in a database, why would Oracle license Oracle Reports separately?
This being said, I find it obvious that either you have not been involved in a major software development project, or that you have been involved in poorly managed projects only. Applying a layer of abstraction between the application and the database is a very good practice and it saves lots of work whenever the application or the database has to be modified.
What do you do if someday for a commercial purpose you must change your database vendor? With your strategy you have to rewrite loads and loads of code, as each vendor has specific SQL features. But if you have an abstraction layer (like EJB) you basically change the drivers and, maybe, update the XML schema.
then next time we try to get rid of this relationship with your ridiculous country, don't come in Montreal a few days before the referendum for a "love-in".
and try to convince your corrupt government not to speed up the processing of immigration in the months before the referendum just to get a few thousand more votes for this Good Ol' Canada like they did in 95.
and while you are at it, pay yourself for the visits of your queen of england which we don't give a damn about, and give us back our share of the money wasted on your 2 Billions $ firearm program which is a big joke.
That's a very interesting point. It's just bad you give none of them reasons.
You'd rather code yourself all the interactions with the storage media instead of trusting a technology that has been specifically designed to understand a data structure and make the best use of it? Then how can you trust a RDBMS to handle the interaction with the filesystem? And then, how could you trust the o/s to deal with the hardware? At some point you have to rely on the software.
I know, it's difficult to let go those lovely queries. But soon SQL will be deprecated, just like COBOL, RPG, Clipper, Fortran, QBasic and others. Either you take the new wave, or you become a "legacy expert" and wish hard that there will be another Y2K-ish scam someday.
So my guess is that either the CFOs feel bad about selling incredibly overpriced products, or they just plain don't understand how the hell they can manage all these crazy contracts.
When Microsoft licensing is the low-cost alternative to your product, there is something terribly wrong.
This is a wrong statement. Data persistence does not have to be bound to the storage technology. Please take time to read about a technology before making so bold statements.
Have a look here.
In the world of J2EE you can use container-managed persistence on entity beans (a flavor of EJB). Such object will handle the interaction with the database for you so you don't have to worry about SQL. And it will also address transaction issues so no need to fiddle with manual rollbacks.
With CMP, instead of fooling around with SQL queries you deal with business objects. Why bother with "update inventory" or "insert into cashRegister" when you can call methods on objects, like inventoryItem.substract() or cashRegister.feed() ? This provides you with a good layer between the application and the database, so whenever one has to change you don't have to mess with both. All you got to do to have all this magic is to keep some XML configuration up to date, and if you have the big bucks you can even get JBuilder to do it for you!
SQL is sooo yesterday. Stop wasting time on technical issues: add business value to your applications instead. And with JBoss you don't even have to pay for a good EJB container. (Of course you'll need a good CPU and a little more RAM, but hey, magic has a price!).
Take this statement: "All rules have an exception".
If this statement is true, then this rule itself must have an exception, in which case the statement is not true. However, if the rule itself is the only one without an exception, then the statement is making itself true by not having an exception.
What is a paradox? A simple demonstration that human logic is flawed.
Great idea: let's take all the networking slang we can find on Google Groups and patch them together in the phoniest post ever made up.
Curiously Netgear has 10/100 Mbps hubs AND 10/100 Mbps switches
You could also use 1-mile category 5 cables wrapped around a gigantic magnet, this would efficiently limit the bandwith of every computer.
Seriously, how the hell this solution can be voted as "3, Informative"??? This is Slashdot, not Fox News!