If you have trouble with concepts, rather than the specifics of how Oracle implements those, you might want to look at e.g. "C.J. Date, An Introduction to Database Systems". It's the "classic" in its field.
Sometimes I wonder how amazingly better than it already is debian would be if the debian developers would spend even half the time they spend flaming each other, and anyone unlucky enough to set his foot in their mailing lists, on actually improving the software.
The differentiator will come out later with more cores. The funky bit about Bechilstienenen's (sp) design is to do with CPU and Memory scaling.
Actually, I highly doubt it. As you might know the Opteron has an integrated on-die memory controller. Ripping that out and replacing it (that is, manufacturing another chip entirely) with something Sun-specific would destroy the x86 economies of scale that is the entire point of this line of servers.
You forgot about Cray. IIRC, they sell systems with hundreds of Opterons...
Well yeah, Cray sells you XT3 systems with up to 30000 Opterons if you fork over enough cash. But those are distributed memory computers. The parent poster fawning over big Sun SPARC:s probably meant shared memory computers. In this class the winner is SGI with currently a maximum of 1024 (or was it 2048) CPU:s running a single Linux kernel.
My university also uses the Sun Messaging server. But we're only about 15000 students, so it's not a huge deal. But it works really well, at least compared to the old system with NFS-mounted mailboxes; there were constant problems with that, and it was overloaded and slow too.
Actually, since what matters is the energy consumption, and this method according to the article delivers 13 MJ/l, it looks very efficient. That's about 1.3 MJ/km.
Compare that to a normal gasoline car that does, say, 7 l/100 km. Gasoline having an energy density of about 45 MJ/l this works out to 3.15 MJ/km.
That is, the hypothesis is that the hydrogen car would be 2.4 times as efficient as the current gasoline car.
And that was the biggest mistake they made with OSX. I mean, by the early 1990:s it was clear that Mach performance sucks. That doesn't mean that microkernels as such are bad, see e.g. L4. Just that Mach sucks. Too bad Avi T. still is in denial.;-)
A traditional monolithic kernel might be "old-fashioned", but they have been proved to work.
1) python multithreaded scalability is limited by the global interpreter lock.
2) With a multi-process model you can use memcached to cache stuff in memory without going via the DB. Additionally, with this architecture horizontal scaling is also possible.
3) A multi-process model is pretty robust. If one process dies, apache restarts it. If your application, or the app framework etc. suffers from a memory leak, you can configure apache to let each process handle only a limited number of requests until it is killed and restarted.
Hydrogen has three common isotopes: protium, deuterium, and tritium. They all have one proton, and zero, one, and two neutrons respectively. Water molecules (H2O) can have any of these isotopes as the hydrogen atoms. When water is made with deuterium and tritium atoms, it is called heavy water. Heavy water is used to regulate fission reactions in nuclear power plants. A heavy water refinery extracts the trace quantities of heavy water molecules from ordinary water.
Actually, the vast majority of commercial nuclear reactors are so-called "light water reactors" because they use, uh, "normal" water. The exception that comes to mind is the Canadian CANDU reactors. CANDU was originally designed to use unenriched uranium, and thus they needed the lower neutron adsorption cross section of heavy water.
So in a way it's a tradeoff. Either you need an enrichment facility, or you need a heavy water plant. Not both.
I remember in high school a younger neighbor of mine got a GI Joe space shuttle. It was a lifting body with a rocket powered and MANNED booster stage. It was not a jet, but it was a liquid fueled space craft on it's own (well, it looked like it was). Anyway, I always thought why did they not build something like that instead of the current design?
I think I have seen some drawings of some booster stage with flyback capability. With todays electronics, there's no need to have it manned though. But anyway, the problem it that the wings add weight. It's just cheaper to put parachutes on the boosters and pick them up where they land.
Anyway, I think the plane format has merit as soon as we figure out how to make more compact and efficient rocket engines.
Rocket engines are plenty compact and efficient already. The problem is that there is only so much thrust you get from chemical fuels. We're already pretty near the limit. Nuclear might be an option, but suffers from political problems.
If they figured out how to make engines with equivalent thrust to the SSME's without the massive fuel requirements, it could work.
Um well, if they could get that to work pretty much any sci-fi scheme would work too. With chemical fuels it won't happen.
IIRC, the Mercury/Gemini/Apollo escape towers were engineered primarily to deal with problems on liftoff. It's not too many minutes into the flight that you're 1000 miles downrange and way out of the breathable atmosphere. What then?
Uh, it's not like the astronauts have to somehow climb up into the tower to use it. They sit in the capsule all the time, the purpose of the tower is to pull the capsule free from the rest of the spacecraft. As the capsule is pressurized and contains life support equipment so that the crew can survive in space as well as the heat shield and parachutes to get the crew safely down from space, I don't see why even a very high altitude abort would be a problem?
Rutan and friends have set up a new company called t/space that's working on a spaceship that would be capable of reaching LEO. It basically looks like a traditional cylindrical rocket with a capsule for 4 guys that is launched from an aircraft. See here.
One neat thing is that they are using a small parachute to turn the rocket vertical after the plane drops it and just before the rocket engine fires. That way they can get rid of wings, saving weight.
The shuttle program is equivalent to saying "These paddleboats just have no future. Let's go back to sails."
I'd rather say that the space shuttle is like a paddleboat before the steam engine was invented. Perhaps it's a good idea, perhaps not, but without a steam engine to turn those paddles there's not much point in it. Sailing usually beats muscle power (rowing or turning the paddles by hand).
The hard things in space flight at our current technology level is getting to orbit and reentry. Putting wings on a space craft is optimizing for the wrong problem. As can be seen by the fact that with a "traditional" capsule on top design the astronauts would have survived a Challenger style accident, and the Columbia accident would never have happened.
Just like we have replaced 4 wheel cars with 5 wheel ones. You know, 4 wheels is so old so its gotta be bad, right?
The evidence seems to be pointing strongly in the direction that a traditional multi-stage rocket is the cheapest and safest way to space with current technology. At some point in the future when we have better materials and much better propulsion some kind of single stage spaceplane might make sense.
Re:Whatever happened to single-stage-to-orbit?
on
NASA's Shuttle Plans
·
· Score: 1
As long as we're no longer trying to send up cargo along with personnel, now might be a good time to revisit single-stage-to-orbit designs such as the Delta Clipper and the Roton.
I don't recall any debris problems with either of these designs, although the leg design seriously needs to be rethought. If you have four legs, a failure of any leg results in disaster (witness the spectacular failure of the Delta Clipper). Six legs, on the other hand, would be far more stable...you could lose any three (provided they're not all adjacent) and still pull off a successful landing.
My gut feeling is that SSTO is based on the same kind of economics as the space shuttle itself. I.e. the admittedly intuitive idea that a reusable spacecraft, a space shuttle if you like, would be oh-so-much cheaper to operate in the long run. Still, history teaches us that the shuttle is the most expensive and among the most risky ways of getting stuff into space ever created.
A SSTO spacecraft would only make these problems worse. And why? You might save a few bucks by having reusable engines and fuel tanks, but that can be done much cheaper by letting the used rocket stages parachute back to earth. Hauling everything to orbit requires a huge amount of energy, compared to a multistage design.
That is not to say that SSTO is a bad concept as such. When we have materials that would allow a reusable spacecraft to be used truly like an airplane as opposed to the huge amount of maintenance work the shuttle undergoes between flights, and when we have better engines (nuclear?), then SSTO might be a more cost effective concept than the traditional multi-stage rockets. I just don't think technology is there yet.
And of course while they are at it they can lock out bootleg Windows licenses forever, win-win for them. And if not outright outlaw Linux, at least make sure only generic whitebox motherboards from Taiwan run it. The Dell and HPs will all be locked to the copy of Windows married to their TCPA module during manufacturing.
I don't think world domination will be that easy for MS. For once, must "larger than mom and pop" businesses reimage their PC:s anyway; they'd scream bloody murder if only the original OEM Windows would work. So if you can put some arbitrary Windows image on a DRM PC, certainly it should then also be possible to install Linux. Also, I guess that at least for servers the Linux market is big enough that vendors cannot afford to let it go down the drain due to some draconian DRM scheme.
Secondly, if MS somehow implements a foolproof anti-piracy scheme, the 3rd world would switch to Linux in a heartbeat. There's no way most 3rd world people can justify paying $100 for a Windows licence. Still MS has as big a market share over there as anywhere else, thanks to piracy. They're counting on those countries getting richer and eventually being able to afford real Windows licenses (at that point the foolproof anti-piracy scheme might make sense to catch those who pirate even though they have lots of disposable income).
Now that they're switching to x86 I guess that they would be able to produce hardware at more competetive prices. Not as cheap as Dell, since Dell is a very streamlined operation which sells huge numbers at small margins; every step of the chain is incredibly lean. No way Apple can match that.
But I don't think they need to match it either. Apple customers pay for the sleek design and hardware that just works, no crappy chips with even crappier drivers etc. By controlling the hardware (and having some sort of official Apple approved peripherals list) they can give their users the same experience that they have given so far on powerpc, but slightly cheaper due to x86 economics. This way they can also sell at slighly higher margins than Dell.
Slashdot Mirror
If you have trouble with concepts, rather than the specifics of how Oracle implements those, you might want to look at e.g. "C.J. Date, An Introduction to Database Systems". It's the "classic" in its field.
Umm, yeah.
Sometimes I wonder how amazingly better than it already is debian would be if the debian developers would spend even half the time they spend flaming each other, and anyone unlucky enough to set his foot in their mailing lists, on actually improving the software.
Actually, AFAIK a lot of crematories use the heat from the cremation for hot water, radiators etc. in the building.
Personally, I don't see anything wrong with that. It's not like the dead care anyway.
The differentiator will come out later with more cores. The funky bit about Bechilstienenen's (sp) design is to do with CPU and Memory scaling.
Actually, I highly doubt it. As you might know the Opteron has an integrated on-die memory controller. Ripping that out and replacing it (that is, manufacturing another chip entirely) with something Sun-specific would destroy the x86 economies of scale that is the entire point of this line of servers.
high-end servers will always need serious superscalar RISC processors.
What does "superscalar RISC" have to do with SMP scalability?
You forgot about Cray. IIRC, they sell systems with hundreds of Opterons...
Well yeah, Cray sells you XT3 systems with up to 30000 Opterons if you fork over enough cash. But those are distributed memory computers. The parent poster fawning over big Sun SPARC:s probably meant shared memory computers. In this class the winner is SGI with currently a maximum of 1024 (or was it 2048) CPU:s running a single Linux kernel.
My university also uses the Sun Messaging server. But we're only about 15000 students, so it's not a huge deal. But it works really well, at least compared to the old system with NFS-mounted mailboxes; there were constant problems with that, and it was overloaded and slow too.
For example, the Cyrus IMAP server supports single instance store using file system storage (Maildir-style IIRC). You don't need a database to do it.
Incidentally, this is how the Cyrus mail server implements its single instance store.
2.5" are coming to the enterprise (Seagate Savio). Expect to see them in consumer gear too in a couple of years.
See pcguide for some reasons why the move to smaller platters is happening.
Actually, since what matters is the energy consumption, and this method according to the article delivers 13 MJ/l, it looks very efficient. That's about 1.3 MJ/km.
Compare that to a normal gasoline car that does, say, 7 l/100 km. Gasoline having an energy density of about 45 MJ/l this works out to 3.15 MJ/km.
That is, the hypothesis is that the hydrogen car would be 2.4 times as efficient as the current gasoline car.
AFAIK OSX uses Mach 2.5.
;-)
And that was the biggest mistake they made with OSX. I mean, by the early 1990:s it was clear that Mach performance sucks. That doesn't mean that microkernels as such are bad, see e.g. L4. Just that Mach sucks. Too bad Avi T. still is in denial.
A traditional monolithic kernel might be "old-fashioned", but they have been proved to work.
Say, the nv driver using the normal 2D apis vs. Xegl doing sw OpenGL?
I'd say the counterargument to this is that:
1) python multithreaded scalability is limited by the global interpreter lock.
2) With a multi-process model you can use memcached to cache stuff in memory without going via the DB. Additionally, with this architecture horizontal scaling is also possible.
3) A multi-process model is pretty robust. If one process dies, apache restarts it. If your application, or the app framework etc. suffers from a memory leak, you can configure apache to let each process handle only a limited number of requests until it is killed and restarted.
At least by looking at his website it seems that he's a class A nutjob. Anti-gravity, faster than light travel and whatnot. ;-)
If the infinite loop is properly parallelized, that is.
Hydrogen has three common isotopes: protium, deuterium, and tritium. They all have one proton, and zero, one, and two neutrons respectively. Water molecules (H2O) can have any of these isotopes as the hydrogen atoms. When water is made with deuterium and tritium atoms, it is called heavy water. Heavy water is used to regulate fission reactions in nuclear power plants. A heavy water refinery extracts the trace quantities of heavy water molecules from ordinary water.
Actually, the vast majority of commercial nuclear reactors are so-called "light water reactors" because they use, uh, "normal" water. The exception that comes to mind is the Canadian CANDU reactors. CANDU was originally designed to use unenriched uranium, and thus they needed the lower neutron adsorption cross section of heavy water.
So in a way it's a tradeoff. Either you need an enrichment facility, or you need a heavy water plant. Not both.
I remember in high school a younger neighbor of mine got a GI Joe space shuttle. It was a lifting body with a rocket powered and MANNED booster stage. It was not a jet, but it was a liquid fueled space craft on it's own (well, it looked like it was). Anyway, I always thought why did they not build something like that instead of the current design?
I think I have seen some drawings of some booster stage with flyback capability. With todays electronics, there's no need to have it manned though. But anyway, the problem it that the wings add weight. It's just cheaper to put parachutes on the boosters and pick them up where they land.
Anyway, I think the plane format has merit as soon as we figure out how to make more compact and efficient rocket engines.
Rocket engines are plenty compact and efficient already. The problem is that there is only so much thrust you get from chemical fuels. We're already pretty near the limit. Nuclear might be an option, but suffers from political problems.
If they figured out how to make engines with equivalent thrust to the SSME's without the massive fuel requirements, it could work.
Um well, if they could get that to work pretty much any sci-fi scheme would work too. With chemical fuels it won't happen.
IIRC, the Mercury/Gemini/Apollo escape towers were engineered primarily to deal with problems on liftoff. It's not too many minutes into the flight that you're 1000 miles downrange and way out of the breathable atmosphere. What then?
Uh, it's not like the astronauts have to somehow climb up into the tower to use it. They sit in the capsule all the time, the purpose of the tower is to pull the capsule free from the rest of the spacecraft. As the capsule is pressurized and contains life support equipment so that the crew can survive in space as well as the heat shield and parachutes to get the crew safely down from space, I don't see why even a very high altitude abort would be a problem?
Rutan and friends have set up a new company called t/space that's working on a spaceship that would be capable of reaching LEO. It basically looks like a traditional cylindrical rocket with a capsule for 4 guys that is launched from an aircraft. See here.
One neat thing is that they are using a small parachute to turn the rocket vertical after the plane drops it and just before the rocket engine fires. That way they can get rid of wings, saving weight.
The shuttle program is equivalent to saying "These paddleboats just have no future. Let's go back to sails."
I'd rather say that the space shuttle is like a paddleboat before the steam engine was invented. Perhaps it's a good idea, perhaps not, but without a steam engine to turn those paddles there's not much point in it. Sailing usually beats muscle power (rowing or turning the paddles by hand).
The hard things in space flight at our current technology level is getting to orbit and reentry. Putting wings on a space craft is optimizing for the wrong problem. As can be seen by the fact that with a "traditional" capsule on top design the astronauts would have survived a Challenger style accident, and the Columbia accident would never have happened.
Just like we have replaced 4 wheel cars with 5 wheel ones. You know, 4 wheels is so old so its gotta be bad, right?
The evidence seems to be pointing strongly in the direction that a traditional multi-stage rocket is the cheapest and safest way to space with current technology. At some point in the future when we have better materials and much better propulsion some kind of single stage spaceplane might make sense.
As long as we're no longer trying to send up cargo along with personnel, now might be a good time to revisit single-stage-to-orbit designs such as the Delta Clipper and the Roton.
I don't recall any debris problems with either of these designs, although the leg design seriously needs to be rethought. If you have four legs, a failure of any leg results in disaster (witness the spectacular failure of the Delta Clipper). Six legs, on the other hand, would be far more stable...you could lose any three (provided they're not all adjacent) and still pull off a successful landing.
My gut feeling is that SSTO is based on the same kind of economics as the space shuttle itself. I.e. the admittedly intuitive idea that a reusable spacecraft, a space shuttle if you like, would be oh-so-much cheaper to operate in the long run. Still, history teaches us that the shuttle is the most expensive and among the most risky ways of getting stuff into space ever created.
A SSTO spacecraft would only make these problems worse. And why? You might save a few bucks by having reusable engines and fuel tanks, but that can be done much cheaper by letting the used rocket stages parachute back to earth. Hauling everything to orbit requires a huge amount of energy, compared to a multistage design.
That is not to say that SSTO is a bad concept as such. When we have materials that would allow a reusable spacecraft to be used truly like an airplane as opposed to the huge amount of maintenance work the shuttle undergoes between flights, and when we have better engines (nuclear?), then SSTO might be a more cost effective concept than the traditional multi-stage rockets. I just don't think technology is there yet.
And of course while they are at it they can lock out bootleg Windows licenses forever, win-win for them. And if not outright outlaw Linux, at least make sure only generic whitebox motherboards from Taiwan run it. The Dell and HPs will all be locked to the copy of Windows married to their TCPA module during manufacturing.
I don't think world domination will be that easy for MS. For once, must "larger than mom and pop" businesses reimage their PC:s anyway; they'd scream bloody murder if only the original OEM Windows would work. So if you can put some arbitrary Windows image on a DRM PC, certainly it should then also be possible to install Linux. Also, I guess that at least for servers the Linux market is big enough that vendors cannot afford to let it go down the drain due to some draconian DRM scheme.
Secondly, if MS somehow implements a foolproof anti-piracy scheme, the 3rd world would switch to Linux in a heartbeat. There's no way most 3rd world people can justify paying $100 for a Windows licence. Still MS has as big a market share over there as anywhere else, thanks to piracy. They're counting on those countries getting richer and eventually being able to afford real Windows licenses (at that point the foolproof anti-piracy scheme might make sense to catch those who pirate even though they have lots of disposable income).
Now that they're switching to x86 I guess that they would be able to produce hardware at more competetive prices. Not as cheap as Dell, since Dell is a very streamlined operation which sells huge numbers at small margins; every step of the chain is incredibly lean. No way Apple can match that.
But I don't think they need to match it either. Apple customers pay for the sleek design and hardware that just works, no crappy chips with even crappier drivers etc. By controlling the hardware (and having some sort of official Apple approved peripherals list) they can give their users the same experience that they have given so far on powerpc, but slightly cheaper due to x86 economics. This way they can also sell at slighly higher margins than Dell.