Indeed. In the case of Sybase for Linux, however, it performs an explicit file buffer flush on every disk write, to ensure that it's written. Of course, this slows it down something awful, but better safe than fast.
That may be, but Linux just can't cut it in the database world. Until 2.4 is stable, then we won't have scaleable SMP or asynch I/O, both of which are critical for a proper database system.
And is Sun kit is so overpriced, why is Sun selling it faster than it can make it?
Doesn't matter how big an army is.... if it can't get across the English Channel and/or North Sea in sufficient numbers, it can't invade, as the Germans in WW2 found out.
Yeah, that makes sense. The one aspect of the whole thing that I have trouble with (always have, probably always will), is the idea of a massless particle having momentum. I've seen the equation, but it's just too bizarre. Similarly, any moving mass particle has a wavelength....
OK, that's fair enough, but I'm still not convinced:-) I can appreciate intellectually that photons have momentum (I recall the equation relating wavelength to momentum), but I can't quite make the leap.
Why does a highly reflective sail help? Surely, if the sail reflects the photons, then the energy is just reflected with it? Doesn't sail have to absorb the energy to gain the momentum? But as someone else pointed out, if it absorbed the energy, it'd just heat up.
heh:) Seriously, though, Agilent's market share is on average something like 70 to 80 percent. When I worked for what is now Agilent 2 years ago, it accounted for 1/6 of HP's total revenue, with something like 1/20 of the personnel.
The HP/Agilent split had nothing to do with separating sales from R&D. What is now Agilent used to the Telecoms & Medical equipment divisions of HP, which, incidentally, was by far and away the most profitable and successful part of the company.
If this quasar is estimated to have formed when the universe was 8% of its current age, then that means the quasar had only 1 billion years to form.
Now, given that a quasar is most likely a super-massive black hole, then it would have need at least one generation of stars to have formed and died, no? And surely 1 billion years is nowhere near long enough for that?
Wasn't there some talk of merging UnixWare with another commerical UNIX (HP-UX, I think). The reasoning was that while the market for UNIX was on the up (at the server, anyway), the "variety" of UNIX platforms was confusing and counter-productive. HP was never one for stubbornly clinging to proprietary tech (embracing IA-64, for eaxmple), and combining their efforts with SCO seemed like a good idea at the time. Anyone know if this is still on the cards?
The Citrix/Terminal Server combo is not really a good example. Winframe used to be a separate product, until MS bought the rights to it and called it Terminal Server. They mucked about with in the process, and the end result isn't very well done. Mind you, I haven't even played with Tarantella, so I can't say whether it's any better or not.
This is where Java clients come into their own, really. Next to no config required on the client, just install the JRE. A properly written Swing GUI java client runs at a useful speed, depending on the the VM you use. With hardware getting faster and cheaper by the week, the overhead associated with java is getting less and less significant. All the code is downloaded from the server on demand, which admittedly is only really useful for intranets. Add to that the far higher productivity in writing these applications compared to C++ (and even VB, when it comes to network apps), and you have a winner.
It's true that administering the server can be a major pain in the arse, but given the choice between haring around fiddling with all those clients, and spending a little more time on a single server, I'd choose the server any day. Apart from anything else, a thin-client software app leaves less for the user to screw^H^H^H^H^Hfiddle with, and gives the admin more control over the application as a whole.
Mark, have you ever tried _administering_ dozens of PCs, each with a complex multi-tier application running on them? When it comes to upgrade time, the thinner the client, the better. Don't confuse thin-client hardware with thin-client software. The former is a niche market. The latter is just common sense.
Slashdot Mirror
That's hardly fair. MySQL is in no way "equivalent" to MS SQL Server, which is in the class above.
That said, you could just as easily use Sybase ASE on linux, which is effectively free, and is a proper RDBMS.
Hmmmm... the Daily Mail is a UK tabloid paper.... need I say more?
Damming the Strait of Gibralter might annoy the large volume of shipping that passes through that gap every day :-)
Indeed. In the case of Sybase for Linux, however, it performs an explicit file buffer flush on every disk write, to ensure that it's written. Of course, this slows it down something awful, but better safe than fast.
Actually, Sun kit _does_ support IDE drives, but only on the lower end workstations.
You missed PA-RISC. HP originally intended to ditch it, but once they realised that IA64 was a turkey, they extended it's life.
The PA-RISC 8500 was the world's fastest CPU when it came out 2 or 3 years ago.
Does 8MB L2 cache mean anything to you?
That may be, but Linux just can't cut it in the database world. Until 2.4 is stable, then we won't have scaleable SMP or asynch I/O, both of which are critical for a proper database system.
And is Sun kit is so overpriced, why is Sun selling it faster than it can make it?
Doesn't matter how big an army is.... if it can't get across the English Channel and/or North Sea in sufficient numbers, it can't invade, as the Germans in WW2 found out.
And listen as the whirring starts....
Yeah, that makes sense. The one aspect of the whole thing that I have trouble with (always have, probably always will), is the idea of a massless particle having momentum. I've seen the equation, but it's just too bizarre. Similarly, any moving mass particle has a wavelength ....
Why does a highly reflective sail help? Surely, if the sail reflects the photons, then the energy is just reflected with it? Doesn't sail have to absorb the energy to gain the momentum? But as someone else pointed out, if it absorbed the energy, it'd just heat up.
Argh
After all, photons have no mass, so where does the momentum come from?
Are any of the sequels as good as the first?
heh :) Seriously, though, Agilent's market share is on average something like 70 to 80 percent. When I worked for what is now Agilent 2 years ago, it accounted for 1/6 of HP's total revenue, with something like 1/20 of the personnel.
The toughest problem has to be the power source. It has to have a very high energy density, and right now, there's no tech really suitable for that.
The HP/Agilent split had nothing to do with separating sales from R&D. What is now Agilent used to the Telecoms & Medical equipment divisions of HP, which, incidentally, was by far and away the most profitable and successful part of the company.
I have a question....
If this quasar is estimated to have formed when the universe was 8% of its current age, then that means the quasar had only 1 billion years to form.
Now, given that a quasar is most likely a super-massive black hole, then it would have need at least one generation of stars to have formed and died, no? And surely 1 billion years is nowhere near long enough for that?
What am I missing?
Is it usable, from a user's perspective?
What the....? Talk about divergence from the thread....
Wasn't there some talk of merging UnixWare with another commerical UNIX (HP-UX, I think). The reasoning was that while the market for UNIX was on the up (at the server, anyway), the "variety" of UNIX platforms was confusing and counter-productive. HP was never one for stubbornly clinging to proprietary tech (embracing IA-64, for eaxmple), and combining their efforts with SCO seemed like a good idea at the time. Anyone know if this is still on the cards?
The Citrix/Terminal Server combo is not really a good example. Winframe used to be a separate product, until MS bought the rights to it and called it Terminal Server. They mucked about with in the process, and the end result isn't very well done. Mind you, I haven't even played with Tarantella, so I can't say whether it's any better or not.
This is where Java clients come into their own, really. Next to no config required on the client, just install the JRE. A properly written Swing GUI java client runs at a useful speed, depending on the the VM you use. With hardware getting faster and cheaper by the week, the overhead associated with java is getting less and less significant. All the code is downloaded from the server on demand, which admittedly is only really useful for intranets. Add to that the far higher productivity in writing these applications compared to C++ (and even VB, when it comes to network apps), and you have a winner.
It's true that administering the server can be a major pain in the arse, but given the choice between haring around fiddling with all those clients, and spending a little more time on a single server, I'd choose the server any day. Apart from anything else, a thin-client software app leaves less for the user to screw^H^H^H^H^Hfiddle with, and gives the admin more control over the application as a whole.
Mark, have you ever tried _administering_ dozens of PCs, each with a complex multi-tier application running on them? When it comes to upgrade time, the thinner the client, the better. Don't confuse thin-client hardware with thin-client software. The former is a niche market. The latter is just common sense.