On Dell's web site you have great flexibility in customizing a system. Except there's no option to delete Windows and save the 30 bucks. You have to jump to another completely different Web site, and maybe you save 30 bucks. Or maybe not. And will your promo code, coupon, or discount code work?
In related news, the State of California also announced that the trial will take place in Cupertino. Podcasts of each day's proceedings will be available from a state Web site.
According to the Washington Post story, there are 21 million households that have analog TVs without satellite or cable (i.e. with only over-the-air reception). (I'm one of them.) Senator Stevens estimates that the government will pay $40 per converter box. Let's assume each of those 21 million households needs two converter boxes (for two TVs, and really the government ought not be paying for more than two). That's $1.68 billion. Add administrative overhead and we're still talking $2 billion or less.
So how does Congress spend $3 billion for a $2 billion "problem"?
It's probably appropriate to mention that IBM Research once had a rival of sorts: Bell Labs. Bell Labs and IBM Research were two of the very few commercial institutions that engaged in basic scientific research -- research that would often yield scientific breakthroughs but much less often commercial success. Now Bell Labs is all but gone, but IBM Research thrives. Thank goodness for IBM Research, and kudos to the IBM managers who still keep the "this quarter" Wall Street monsters at bay in order to spend the billions it takes for science.
Assuming each of your blade servers has two CPUs -- not a bad assumption -- Microsoft just quintupled your software bill, didn't they? All the more reason to switch to Linux, Linux software, and processors that can do better than a 5 to 1 VM ratio. (Mainframe Linux comes to mind.)
Most IBM software is priced per CPU. And everything after that is in the customer's favor.
If it's a dual core CPU you pay for one CPU, not two (unlike, say, Oracle). If you use virtualization software (like z/VM, LPARs, Virtual PC, or VMware) you only pay for the number of CPUs that the software actually executes on. If that means you run 300 instances of DB2 for Linux on a single Linux mainframe CPU running z/VM, you pay for one CPU, not 300. Unlike Microsoft. If you want to switch from DB2 for Windows to DB2 for Linux (on the mainframe or anywhere else), fine -- the processor licenses are cross-platform. Don't pay again.
The main reason corporate customers run virtual machine technology is so they can consolidate the ridiculous numbers of test and development servers which cost a fortune. Under IBM's pricing policy that's encouraged, and they can get their costs under control. Under Microsoft's new policy it'll cost those businesses more if they use virtualization to any significant degree.
You should consider migrating those Unisys mainframes to an IBM mainframe. The IBM mainframe is thoroughly modernized and thriving (Linux!), and it's the one platform that has the characteristics you're going to need. It's also the lowest total cost platform for organizations that are of any scale (i.e. not the corner hardware store).
An IBM mainframe is also much more likely to have the tools and resources needed to pull off a successful migration. And there are many, many former Unisys shops that have made the switch, so you won't be the first.
Your Commonwealth's taxpayers will be much happier if you cut your losses now and, unlike Unisys, IBM will actually want to help you get off Unisys.
Re: DB2, no, in a word. Most CRM software (by far) is in-house developed, and much of that is DB2-based. And there's a not-so-small CRM company by the name of SAP which will sell a lot more if Siebel did anything to mess with DB2 support. (SAP is already salivating, I'd bet.) And there's that DB2 z/OS product where huge amounts of enterprise data are kept. Oracle z/OS blows chunks (though their zSeries Linux product is good). If you don't support DB2 z/OS you're screwed in just about any enterprise environment. And IBM will happily sell you hardware and services regardless.
Actually, IBM does fine, but the big loser is probably Microsoft. Oracle believes in both Linux and J2EE, two concepts that are not Microsoft's. Siebel had previously announced they'd support both.NET and J2EE in their next big version. The.NET version is bound to disappear completely now. Siebel also announced that they'd support WebSphere Application Server as their J2EE runtime of choice. I think that'll continue -- it's hard to be enterprise J2EE without supporting the #1 runtime -- but I suspect Oracle will also allow using Oracle's own J2EE runtime as an alternative. Siebel also promised they'd move away from their Internet Explorer-only user interface. They've already started to do that, but you can bet Oracle will continue that trend and make sure that Siebel works great with Firefox, et. al.
Is anyone working on an open source OpenDocument import/export filter for Microsoft Office? Just like Firefox for Windows (as a transition vehicle from Internet Explorer), that'd help start to wean people away from Microsoft Office.
If only the NYT reporter used Google! According to this article, here's what BAE used:
The BAE design includes a number of high-tech components. It calls for 300 486-class computers distributed in eight control rooms, a Raima Corp. database running on a Netframe Systems fault-tolerant NF250 server, a high-speed fiber-optic ethernet network, 14 million feet of wiring, 56 laser arrays, 400 frequency readers, 22 miles of track, 6 miles of conveyor belts, 3,100 standard telecars, 450 oversized telecars, 10,000 motors, and 92 PLCs to control motors and track switches.
Let's make some educated guesses here. How many PCs? 300? Good grief! Later in the article it says the PCs were running OS/2. So what? This is just bad architecture, regardless of OS. So many parts, so many points of potential failure. And the NetFrame Systems "fault tolerant" server is simply...a glorified PC. (It's X86, ~300 MHz P2, and likely running Windows NT, according to other sources.) This article has more on the sad fate of NetFrame.
There's nothing even close to a mainframe computer in this baggage handling system. The New York Times sucks again!
You're exactly right. I really wish the NYT reporter would have done a better job figuring out why the baggage system didn't work. Instead we're left guessing, with vague anecdotes about carts tipping over and barcodes not being properly scanned -- all of which has nothing whatsoever with the computer in the back office.
By the way, mainframes are at the heart of every major package shipper's operations, tracking every little bit of package-related data. Not so surprising: it's a job that must run reliably, without downtime. Mainframes do that well.
And if the mainframe was to blame, why didn't United (or the airport or whomever) just drop in one of those wonderful Peecees to do the job? That would have fixed everything, right? Obviously no, that wasn't the problem.
CERN had the first Web site to integrate with a backend system. According to history, back in 1990 CERN developed a Web site that provided dynamic access to a VM FIND application. Thus the world's first Web application integration project provided Web access to an IBM mainframe application. (It's also true that the world's first Web server outside Europe was installed on Stanford's IBM mainframe.)
Seriously, mainframes are so cool. And they offer patent protection, too.
Let's try this again...:-) The E20 is not a CHRP system (evidently) but should be PReP. Thus it will run AIX up through 5.1 (but not 5.2 or 5.3), and so its support days are at best limited. Debian Linux *does* support the PReP systems.
Just to give you an idea of the age of this server, the E20 debuted in 1995 (at 100 MHz). There were some processor upgrade options released later. Maximum system memory is 512 MB, I believe.
Ooops, I take that back. AIX 5.3 supports a 32-bit multiprocessor kernel that is compatible with all CHRP systems, including uniprocessor models. I misread that. So it looks like the buyer did even better than I thought.
Debian runs on CHRP systems, so the E20 would make a pretty good Debian Linux system.
The E20 would be a 32-bit PowerPC-based (604) server of the 100 MHz to 233 MHz variety (probably 100 MHz). Hard disk sizes would likely be in the 9 GB per disk range. Memory would be around 256 MB or perhaps more if upgraded. But the real limiting factor is that AIX support for the 32-bit hardware is coming to a close. (The 64-bit hardware has been available for quite some time now, and the latest AIX doesn't even run on 32-bit hardware.)
These servers could be nicely rehabilitated with Linux, however. In fact, they might make excellent testbeds for developers who wish to compile for Linux on POWER (in lowest common denominator fashion). And IBM hardware is deservedly respected for its quality, and these are server-class machines (unlike, say, a PowerPC 604-based Macintosh). So the buyer did very well, IMHO.
Slashdot Mirror
On Dell's web site you have great flexibility in customizing a system. Except there's no option to delete Windows and save the 30 bucks. You have to jump to another completely different Web site, and maybe you save 30 bucks. Or maybe not. And will your promo code, coupon, or discount code work?
In related news, the State of California also announced that the trial will take place in Cupertino. Podcasts of each day's proceedings will be available from a state Web site.
So how does Congress spend $3 billion for a $2 billion "problem"?
It's probably appropriate to mention that IBM Research once had a rival of sorts: Bell Labs. Bell Labs and IBM Research were two of the very few commercial institutions that engaged in basic scientific research -- research that would often yield scientific breakthroughs but much less often commercial success. Now Bell Labs is all but gone, but IBM Research thrives. Thank goodness for IBM Research, and kudos to the IBM managers who still keep the "this quarter" Wall Street monsters at bay in order to spend the billions it takes for science.
Assuming each of your blade servers has two CPUs -- not a bad assumption -- Microsoft just quintupled your software bill, didn't they? All the more reason to switch to Linux, Linux software, and processors that can do better than a 5 to 1 VM ratio. (Mainframe Linux comes to mind.)
Most IBM software is priced per CPU. And everything after that is in the customer's favor. If it's a dual core CPU you pay for one CPU, not two (unlike, say, Oracle). If you use virtualization software (like z/VM, LPARs, Virtual PC, or VMware) you only pay for the number of CPUs that the software actually executes on. If that means you run 300 instances of DB2 for Linux on a single Linux mainframe CPU running z/VM, you pay for one CPU, not 300. Unlike Microsoft. If you want to switch from DB2 for Windows to DB2 for Linux (on the mainframe or anywhere else), fine -- the processor licenses are cross-platform. Don't pay again. The main reason corporate customers run virtual machine technology is so they can consolidate the ridiculous numbers of test and development servers which cost a fortune. Under IBM's pricing policy that's encouraged, and they can get their costs under control. Under Microsoft's new policy it'll cost those businesses more if they use virtualization to any significant degree.
You should consider migrating those Unisys mainframes to an IBM mainframe. The IBM mainframe is thoroughly modernized and thriving (Linux!), and it's the one platform that has the characteristics you're going to need. It's also the lowest total cost platform for organizations that are of any scale (i.e. not the corner hardware store). An IBM mainframe is also much more likely to have the tools and resources needed to pull off a successful migration. And there are many, many former Unisys shops that have made the switch, so you won't be the first. Your Commonwealth's taxpayers will be much happier if you cut your losses now and, unlike Unisys, IBM will actually want to help you get off Unisys.
Actually, IBM does fine, but the big loser is probably Microsoft. Oracle believes in both Linux and J2EE, two concepts that are not Microsoft's. Siebel had previously announced they'd support both .NET and J2EE in their next big version. The .NET version is bound to disappear completely now. Siebel also announced that they'd support WebSphere Application Server as their J2EE runtime of choice. I think that'll continue -- it's hard to be enterprise J2EE without supporting the #1 runtime -- but I suspect Oracle will also allow using Oracle's own J2EE runtime as an alternative. Siebel also promised they'd move away from their Internet Explorer-only user interface. They've already started to do that, but you can bet Oracle will continue that trend and make sure that Siebel works great with Firefox, et. al.
Is anyone working on an open source OpenDocument import/export filter for Microsoft Office? Just like Firefox for Windows (as a transition vehicle from Internet Explorer), that'd help start to wean people away from Microsoft Office.
The BAE design includes a number of high-tech components. It calls for 300 486-class computers distributed in eight control rooms, a Raima Corp. database running on a Netframe Systems fault-tolerant NF250 server, a high-speed fiber-optic ethernet network, 14 million feet of wiring, 56 laser arrays, 400 frequency readers, 22 miles of track, 6 miles of conveyor belts, 3,100 standard telecars, 450 oversized telecars, 10,000 motors, and 92 PLCs to control motors and track switches.
Let's make some educated guesses here. How many PCs? 300? Good grief! Later in the article it says the PCs were running OS/2. So what? This is just bad architecture, regardless of OS. So many parts, so many points of potential failure. And the NetFrame Systems "fault tolerant" server is simply...a glorified PC. (It's X86, ~300 MHz P2, and likely running Windows NT, according to other sources.) This article has more on the sad fate of NetFrame.
There's nothing even close to a mainframe computer in this baggage handling system. The New York Times sucks again!
You're exactly right. I really wish the NYT reporter would have done a better job figuring out why the baggage system didn't work. Instead we're left guessing, with vague anecdotes about carts tipping over and barcodes not being properly scanned -- all of which has nothing whatsoever with the computer in the back office. By the way, mainframes are at the heart of every major package shipper's operations, tracking every little bit of package-related data. Not so surprising: it's a job that must run reliably, without downtime. Mainframes do that well. And if the mainframe was to blame, why didn't United (or the airport or whomever) just drop in one of those wonderful Peecees to do the job? That would have fixed everything, right? Obviously no, that wasn't the problem.
Seriously, mainframes are so cool. And they offer patent protection, too.
Just to give you an idea of the age of this server, the E20 debuted in 1995 (at 100 MHz). There were some processor upgrade options released later. Maximum system memory is 512 MB, I believe.
Debian runs on CHRP systems, so the E20 would make a pretty good Debian Linux system.
These servers could be nicely rehabilitated with Linux, however. In fact, they might make excellent testbeds for developers who wish to compile for Linux on POWER (in lowest common denominator fashion). And IBM hardware is deservedly respected for its quality, and these are server-class machines (unlike, say, a PowerPC 604-based Macintosh). So the buyer did very well, IMHO.