Ahh.. but we're not talking about "just a S/390" now. The system you describe is certainly going to cost about the same as Hotmails clustered systems. And 99.999 is a lot worse than 100%, which a clustered system can provide (since it's not running in the same physical hardware).
What happens if a meteor or airplane crash takes out your central data center? Clustered systems can exist in multiple locations.
What happens if your Power and backup fails to your ONE clustered device? With arrays, your power is likely coming from multiple sources.
Apparently you don't understand the concept of "discardable" memory. This is memory that is being used, simply because nobody else is using it. Things like disk caches are good examples of this.
If you only have 3 megs of memory free, and you need 10. Discard 7 megs of the least recently used disk cache (which is what the OS will do) and free it up. No big deal. Things aren't swapped to disk when there is discardable memory available.
Hmm.. what happens when you install a MacOS driver in Linux? I'm sure someone could figure out a way to do so, but the point of the matter is that you're running a driver for a different OS (or OS version). Drivers are the achiles heel of any OS, including Linux. X Servers can crash your machine, hard disk drivers can kernel panic you.
Keyboard and Mouse are not controlled by the same driver, though they are closely related and an error in one can screw up the other (both use the same input controller). They may both exist in the same file, but that doesn't mean it's the same driver.
Oh, give me a break. In the absence of a real, credible source that counters the statement, the only thing you can do is take them at their word.
I guess you should just believe the opposite of everything they say right? They say (at the time) it ran on Solaris and FreeBSD. Guess that must be a lie as well, right? I guess we shouldn't believe them that they're currently migrating to Windows 2000.
Selectivly choosing what you want to believe, based only on whether it supports your side is what zealotry is all about.
There are no facts to support a migration to NT in 1998. On top of that, the same magazine that posted the original article posted this article a few weeks later, which talks about how MS has done major and massive overhauls on Hotmail (supposedly, while also doing this massive migration which failed). How exactly does MS get all those overhauls done at the same time they're migrating? There simply isn't enough manpower to do all that.
Oh, you've got to be kidding me. I post a genuine article with real facts (which were requested by the poster of the article) and I get moderated to a 0 with a Flamebait comment.
Also, all it takes is half a brain to think about this. If MS supposedly did the conversion right after buying Hotmail, how could they have?
It certainly takes more than a few months to convert a project as massive as Hotmail over to a new architecture. Testing alone would take 6-8 months. Things like this don't "just happen". They're not using off the shelf software for the Solaris and BSD front ends, it's highly modified. And they have to completely rewrite all the dynamic content to be ASP as well.
Any claim of a conversion shortly after acquiring Hotmail is ludicrous from the standpoint of the massive amount of work such a conversion would take. And work doesn't get done overnight. It takes time. Lots of it. Imagine the effort it would take just to move Slashdot to IIS? And Slashdot (while a technical marvel in and of itself) isn't a fraction of what Hotmail is.
Get over it. It didn't happen. And stop spreading the rumor that it did.
Slashdot Mirror
Ahh.. but we're not talking about "just a S/390" now. The system you describe is certainly going to cost about the same as Hotmails clustered systems. And 99.999 is a lot worse than 100%, which a clustered system can provide (since it's not running in the same physical hardware).
What happens if a meteor or airplane crash takes out your central data center? Clustered systems can exist in multiple locations.
What happens if your Power and backup fails to your ONE clustered device? With arrays, your power is likely coming from multiple sources.
Apparently you don't understand the concept of "discardable" memory. This is memory that is being used, simply because nobody else is using it. Things like disk caches are good examples of this.
If you only have 3 megs of memory free, and you need 10. Discard 7 megs of the least recently used disk cache (which is what the OS will do) and free it up. No big deal. Things aren't swapped to disk when there is discardable memory available.
Hmm.. what happens when you install a MacOS driver in Linux? I'm sure someone could figure out a way to do so, but the point of the matter is that you're running a driver for a different OS (or OS version). Drivers are the achiles heel of any OS, including Linux. X Servers can crash your machine, hard disk drivers can kernel panic you.
Keyboard and Mouse are not controlled by the same driver, though they are closely related and an error in one can screw up the other (both use the same input controller). They may both exist in the same file, but that doesn't mean it's the same driver.
I agree, these things should be unrelated.
Tell me, what happens when that ONE S/390 dies from a hardware or other failure?
That's the whole point of clusters. Redundancy.
You can't help but question?
Why? Microsoft is running it's operations on it's products. Apart from the PR reasons, there are good solid technical reasons to do this.
It makes their products more robust. They have to live with their own bugs, and make damn sure they get fixed.
Oh, give me a break. In the absence of a real, credible source that counters the statement, the only thing you can do is take them at their word.
I guess you should just believe the opposite of everything they say right? They say (at the time) it ran on Solaris and FreeBSD. Guess that must be a lie as well, right? I guess we shouldn't believe them that they're currently migrating to Windows 2000.
Selectivly choosing what you want to believe, based only on whether it supports your side is what zealotry is all about.
There are no facts to support a migration to NT in 1998. On top of that, the same magazine that posted the original article posted this article a few weeks later, which talks about how MS has done major and massive overhauls on Hotmail (supposedly, while also doing this massive migration which failed). How exactly does MS get all those overhauls done at the same time they're migrating? There simply isn't enough manpower to do all that.
Oh, you've got to be kidding me. I post a genuine article with real facts (which were requested by the poster of the article) and I get moderated to a 0 with a Flamebait comment.
Does Slashdot want facts or not?
Also, all it takes is half a brain to think about this. If MS supposedly did the conversion right after buying Hotmail, how could they have?
It certainly takes more than a few months to convert a project as massive as Hotmail over to a new architecture. Testing alone would take 6-8 months. Things like this don't "just happen". They're not using off the shelf software for the Solaris and BSD front ends, it's highly modified. And they have to completely rewrite all the dynamic content to be ASP as well.
Any claim of a conversion shortly after acquiring Hotmail is ludicrous from the standpoint of the massive amount of work such a conversion would take. And work doesn't get done overnight. It takes time. Lots of it. Imagine the effort it would take just to move Slashdot to IIS? And Slashdot (while a technical marvel in and of itself) isn't a fraction of what Hotmail is.
Get over it. It didn't happen. And stop spreading the rumor that it did.