I beg to differ on some points. I am on a project that migrated from a mainframe to distributed unix servers. While licensing costs did go down, and development staff stayed approximately the same the same, productivity went down due to the more complex environment. The infrastucture costs increased by a factor of 3 (from 1 mainframe to 11 large unix servers) and support staff went up by a factor of 4 (to administer all those new boxes). System response times and up-time went down. A properly run mainframe literally has no unplanned downtime and is very, very fast with business applications.
Overall total costs increased somewhat. But since licensing comes out of one department's budget, sysadmins and operator/schedulers are out of another division's budget, while development comes out of yet another budget, the increase in overall costs is not visible to the suits.
We in the trenches know the truth, but the PHB's are not about to go to the board and tell them that this changeover they pushed for is costing at least $350,000 more a year. It wouldn't be a career enhancing move.
It all depends on the scale of the system. A cluster of smaller machines with SAN to handle the size of the databases we use and the number of simultanious users, plus the data archiving requirements we have to have in place, would literally fill a warehouse and be more even more expensive to run.
The mainframe would be hard to replace by clusters. The raw O/I volume alone would saturate networks.
The busses between components are fiberoptic for the bandwith. While your cluster could approach the reliability of a mainframe, you would need far more staffing, infrastructure, and real estate to operate it. A modern mainframe consists of 2 refrigerator sized machines (one is the computer, the other is the hard drives, or DASD's) and would only require 3 or 4 operators and and 2 or 3 system guys to run.
FYI, at one utility that I worked at 6 years ago running a small mainframe would archive off 1.5 terabytes of data a day.
Mainframes do not have raw processing power, but there is nothing else that processes the sheer volume that they do.
Once 10 people have done this then the market is saturated and the pay goes to near minimum wage. And what are the other 50,000 unemployed mainframers supposed to do?
Don't I wish, a lot of mainframe jobs have been offshored. Unless you have a lot of experience in integrating web, unix, and mainframes in hybred systems, the pay isn't any better than what it was 10 yrs ago. I know a fair number of mainframe coders who were laid off and are now doing other things, not because they want to, but because the only jobs they could find were short term temp positions, and they would have to move every 6 months to stay working. The cost to their families was too high, and they simply couldn't afford to uproot and move every few months. I've managed to stay employed, but I've worked for 4 companies in 6 years. I've gone back to school to get another degree, this time in accounting, because I to am planning to get out of the field because I need a reliable paycheck.
Slashdot Mirror
I've actually done things like this in a previous life, it's not for the faint of heart. Ex nuke, now just a coder
I beg to differ on some points. I am on a project that migrated from a mainframe to distributed unix servers. While licensing costs did go down, and development staff stayed approximately the same the same, productivity went down due to the more complex environment. The infrastucture costs increased by a factor of 3 (from 1 mainframe to 11 large unix servers) and support staff went up by a factor of 4 (to administer all those new boxes). System response times and up-time went down. A properly run mainframe literally has no unplanned downtime and is very, very fast with business applications. Overall total costs increased somewhat. But since licensing comes out of one department's budget, sysadmins and operator/schedulers are out of another division's budget, while development comes out of yet another budget, the increase in overall costs is not visible to the suits. We in the trenches know the truth, but the PHB's are not about to go to the board and tell them that this changeover they pushed for is costing at least $350,000 more a year. It wouldn't be a career enhancing move. It all depends on the scale of the system. A cluster of smaller machines with SAN to handle the size of the databases we use and the number of simultanious users, plus the data archiving requirements we have to have in place, would literally fill a warehouse and be more even more expensive to run.
The mainframe would be hard to replace by clusters. The raw O/I volume alone would saturate networks. The busses between components are fiberoptic for the bandwith. While your cluster could approach the reliability of a mainframe, you would need far more staffing, infrastructure, and real estate to operate it. A modern mainframe consists of 2 refrigerator sized machines (one is the computer, the other is the hard drives, or DASD's) and would only require 3 or 4 operators and and 2 or 3 system guys to run. FYI, at one utility that I worked at 6 years ago running a small mainframe would archive off 1.5 terabytes of data a day. Mainframes do not have raw processing power, but there is nothing else that processes the sheer volume that they do.
Once 10 people have done this then the market is saturated and the pay goes to near minimum wage. And what are the other 50,000 unemployed mainframers supposed to do?
Don't I wish, a lot of mainframe jobs have been offshored. Unless you have a lot of experience in integrating web, unix, and mainframes in hybred systems, the pay isn't any better than what it was 10 yrs ago. I know a fair number of mainframe coders who were laid off and are now doing other things, not because they want to, but because the only jobs they could find were short term temp positions, and they would have to move every 6 months to stay working. The cost to their families was too high, and they simply couldn't afford to uproot and move every few months. I've managed to stay employed, but I've worked for 4 companies in 6 years.
I've gone back to school to get another degree, this time in accounting, because I to am planning to get out of the field because I need a reliable paycheck.