It's is WDM, but instead of multiple low frequency lasers firing at once, they've got a single high frequency laser firing multiple wavelengths. They've taken the lead from other high-speed data busses like IDE and SCSI in transitioning from parallel to serial, as (I presume) cross-talk must become an issue as speeds rise.
Ryanair are the cheapest of the cheap when it comes to airlines. You have to pay for your coffee and the stewardesses are ugly. They make their money on the stuff you buy on board, and more importantly what they can sell you when you're on the www.ryanair.com site. They charge for bag check-in and "priority" boarding, and get kick-backs (I presume) from their "partners" who'll sell you car rental, insurance and all that good stuff on the site. People who buy from these external websites don't see any of this stuff, so Ryanair actually lose money, so all those people who say that Ryanair are turning away "good" business are wrong. By killing off the 3rd party websites, Ryanair are actually saving money.
... how the hell do you manage 40+ storage servers, providing disk to $X application servers in the datacenter? I've got application servers running Linux, AIX & Windows in clusters, failover setups, and behind load balancers, so dealing with so many storage nodes will hurt. I could stick everything behind NAS heads, but then I've got a significant performance bottleneck. I could/should RAID 5 each box, but then I've no redundancy if a PSU/CPU/controller/NIC blows, so I think what I'd have to do is run each node RAID 0, with each node functioning as a "disk" of a AoE RAID 5 set managed by the application server. That's no big deal with a low number of application servers, or where each RAID is only accessed by a single server, but it's unworkable if I have to coordinate disk additions and deletions and rebuilds across multiple application servers, isn't it? I guess what I'm asking is "lots of disks are fine, but what management tools are available to make it all work in the real world?"
Ease of Use isn't a major factor in the 5000 user industrial setting anyway. Company IT resources are aimed at enabling the users to use specific tools. In the case of the example, the application was Peoplesoft. The interface is the same on both platforms, therefore the application will be just as easy to use on both platforms. In the example given, the desktop/menu on X or Windows would just have three icons, Peoplesoft, an office suite, and an email client!
It doesn't matter who's command line is easier to learn or more powerfull, or whose menus are more intuitive, because the business user doesn't need these things, and an IT department doesn't want to have to support those features.
The article's main thrust is on cost of ownership and reliability, and while I consider the Windows costs to be slightly inflated, the author makes his arguement based on these factors, which are the ones that matter in a business.
Slashdot Mirror
It's is WDM, but instead of multiple low frequency lasers firing at once, they've got a single high frequency laser firing multiple wavelengths. They've taken the lead from other high-speed data busses like IDE and SCSI in transitioning from parallel to serial, as (I presume) cross-talk must become an issue as speeds rise.
Ryanair are the cheapest of the cheap when it comes to airlines. You have to pay for your coffee and the stewardesses are ugly. They make their money on the stuff you buy on board, and more importantly what they can sell you when you're on the www.ryanair.com site. They charge for bag check-in and "priority" boarding, and get kick-backs (I presume) from their "partners" who'll sell you car rental, insurance and all that good stuff on the site. People who buy from these external websites don't see any of this stuff, so Ryanair actually lose money, so all those people who say that Ryanair are turning away "good" business are wrong. By killing off the 3rd party websites, Ryanair are actually saving money.
... how the hell do you manage 40+ storage servers, providing disk to $X application servers in the datacenter? I've got application servers running Linux, AIX & Windows in clusters, failover setups, and behind load balancers, so dealing with so many storage nodes will hurt. I could stick everything behind NAS heads, but then I've got a significant performance bottleneck. I could/should RAID 5 each box, but then I've no redundancy if a PSU/CPU/controller/NIC blows, so I think what I'd have to do is run each node RAID 0, with each node functioning as a "disk" of a AoE RAID 5 set managed by the application server. That's no big deal with a low number of application servers, or where each RAID is only accessed by a single server, but it's unworkable if I have to coordinate disk additions and deletions and rebuilds across multiple application servers, isn't it? I guess what I'm asking is " lots of disks are fine, but what management tools are available to make it all work in the real world? "
It doesn't matter who's command line is easier to learn or more powerfull, or whose menus are more intuitive, because the business user doesn't need these things, and an IT department doesn't want to have to support those features.
The article's main thrust is on cost of ownership and reliability, and while I consider the Windows costs to be slightly inflated, the author makes his arguement based on these factors, which are the ones that matter in a business.
Regards,
- Tycho