It could be useful for other things. For example, doing tape backups over the network. A full backup of a 50GB drive over the network is going to take two hours (or longer). During that time your server's hard disk is going to be thrashing like crazy and performance will be shot to hell. But if you've got two disks, then you just "break" the array (making one of the disks readonly) and backup from just one disk. Minimal performance hit, you get the benefits of RAID1 (and snapshots), and it's a pretty simple system so there's not too much that can go wrong. We have a couple hundred servers at work that use this approach and it's amazingly convenient and problem-free. A lot less flaky than most RAID hardware too, sad to say. (I'm looking in your direction, Mylex and LSI.)
I can't speak for how OSX does things, but I agree that keyboard shortcuts in Windows are pretty solid. My main beef is that so many user apps don't implement them properly or at all.
It's possible to control every aspect of a Windows computer (except MMC, dammit) from the keyboard. Except for third-party apps, which are often broken - but you can't blame MS for that. Most keyboard shortcuts have been Windows-standard since the 3.0 days, it's not MS's fault that so many developers are too lazy to implement them.
I would like visual feedback a la MacOS (or Firefox, for that matter) for typeahead find in folders though. But oh well, I type fast enough for it not to be a problem.
Also, are the clipboard shortcuts really different? I seem to remember they're Option+ZXCV, which is not terribly different from Ctrl+ZXCV. It's some vendor-agnostic usability standard; I remember IBM making a big deal of it back in the OS/2 days, when Windows (3.0) wanted you to use Ctrl/Shift+Insert and Shift+Delete for copy, paste, and cut, respectively. (They still work, try them out.)
I think you are misunderstanding. The reviewer is talking about Intel chipsets, not Intel processors. Because Intel doesn't make chipsets for AMD-CPU systems, there's no direct comparison. If there were, he's arguing, people would be less excited about NVIDIA's chipsets.
Incidentally:
These days, AMD is the singlemost important chip maker on the planet, second only to Intel.
It's hard to be both most important and second-most important at the same time. Yet this is apparently a feat both AMD and Intel have managed. I guess this is a byproduct of their research into quantum computers. ("Alright! We're on top! Oh shit, I just changed our importance by measuring it.")
Sure, they have a choice. They can take their ball and go home. "There are levels of survival we are prepared to accept." Of course, it would never go that far. MS would threaten it, maybe stop selling Windows in Europe for a few weeks, then both sides would panic and split the difference.
Or MS can say "The fines for noncompliance will be lower than the cost of damage to our business if we do comply. So we're going to eat the fines." Even if that were true, eventually the EU would either crank up the fines or simply bar MS from selling Windows as-is, in which case we're back to option 1.
"Not proof" means the absence of proof. "Contra-proof positive" means there is proof, but it's directly contrary to what the original author claims.
Say I claim that the sky is red and offer evidence to that effect. If my evidence is inadequate, then it's not proof. If my evidence proves that the sky is definitely not red, then it's contra-proof: it proves the opposite of what I am claiming.
It would be less awkward to say "proof negative," but the contra- prefix is common in philosophical circles where this sort of fine distinction is usually relevant.
So NASA decides a robot mission to rescue HST is not feasable, but the same time they develope this kind of mission for other purposes?
That doesn't really surprise me. Satellites aren't built to be repaired. If they get damaged too badly they're dropped into the ocean. Launching the Shuttle into space is simply too expensive unless the satellite is really valuable. With the Columbia disaster, Shuttle launches just became too expensive to repair any satellites, and it has nothing to do with money.
I would expect that as these satellite-repairing bots become more advanced we will see newer satellites which are designed to allow easier repairs.
I wonder how many infrared telescopes build for the secret service are already monitoring us...
I love how asserting any paranoid theory on Slashdot gets you modded insightful.
On an unrelated topic, there is a herd of purple dancing elephants who live in my closet, but light makes them invisible. Sometimes at night, in the dark, I can see them laughing at me. My friends say it's just the outline of my clothes being blown by the air conditioner, but I know they're lying. It just proves that my friends are all on the payroll of the elephants. We're fast approaching a world run by a shadowy heliotrope proboscidean ecdysiast cabal operating out of my bedroom. It may already be too late to stop it.
It depends on the state, I suspect. For a while, my parents both lived in Massachusetts while my mom worked there and my dad worked just across the border in Connecticut. Because they were married and filing jointly, they had to pay both state taxes on their (full!) income. (Actually I believe CT didn't have a state income tax at the time, but still...) Naturally we moved to CT soon after. Insert Taxachusetts jokes here, because they're surely on-topic.
These sorts of problems are very common in the Northeast where the states are so small. On the one hand, the ruling seems almost fair: you're getting the benefits of NY (mass transit, cops, etc) but not paying a dime in taxes. But on the other hand, any taxes an employer pays are effectively taxes levied on all its employees (and customers!), so I'm not sure that really holds water. (This is that "double taxation" that some flat-tax advocates talk about.)
You could put in a clause stating that you grant an irrevocable, non-transferrable, royalty-free license to use any patents owned by the copyright holder in the source code. Probably "irrevocable" would need to be lawyered up, since the GPL permits you to use (but not distribute) a program even if you don't agree to the license. (So you would want to revoke the patent license in that case.)
Anyway, that would let open source developers use code contributed by IBM and other big IP holders free of worry about being sued some day. Even if IBM (for example) contributed code using patents it doesn't own (such as SCO), developers would have an extremely good defense. "IBM and its horde of patent lawyers told me I could use it." Continuing with my hypothetical situation, it would also apply to people who use the software, so long as they have agreed to the GPL, and so it would protect Linux users and not just developers.
That's my thinking. But neither I nor anyone else who has posted to this story has any clue what will really be in GPLv3, so... you know, don't be surprised if I turn out to be totally wrong.
As someone who does a fair bit of plugging together home electronics equipment, it's also important to get quality connectors and shielding.
The former is important because you will need to unplug those bastards occasionally, and I have some Radio Shack cables that I have to disconnect with a wrench and significant amounts of muscle. I'm concerned about literally breaking the connector from the amount of force I have to apply. From the other direction, if they come off too easily you will have to tape them to the TV or something, which doesn't impair signal quality but is annoying.
The latter is important not because of interference or anything, but because stuff will happen to the cables. You'll vacuum it, your pets will chew on it, you'll get crap on it, etc. It's a pain in the ass to get behind your TV to replace frayed or broken cables, so you should try to minimize it.
Note that neither of these considerations imply that you should buy Monster cables, just that you should avoid the extremely inexpensive (read: cheap) ones. So instead of buying the $30 Monster cables or the $5 generic cables, buy the $15 name-brand-but-not-Monster cables. Belkin seems like a pretty good target for that, and you could in fact do better if you find a place that lets you return cables. (You obviously have to open up the package to try them out, and some places don't want to take them back after you do that.)
I do always laugh when I see the rusted-out '85 Honda Civic with the brand new shiny 20" rims and chrome exhaust tips. "You just spent $2500 on mods, and a Civic is the best you can do?"
Let's do unto others, here. Most Slashdotters prefer open source operating systems, but a lot of us are forced to use Windows because we need some program that only runs there. So we all beg developers to release Linux versions of popular programs and dream of a day when one app runs on all operating systems.
Since we know the pain of being forced to use an OS we dislike, let's try to be the Better Man here. If the tables someday turn, let's continue to encourage that apps get ported to every OS under the sun. Not because it helps us (it wouldn't, in this hypothetical world), but because it's the right thing to do.
By the same token, I support the PSQL folks porting their excellent software to Windows. Let those who, for whatever reason, want to run Windows make that choice themselves and still be able to run PSQL.
Well, remember that MS's products are used on hundreds of millions of computers worldwide, and after the OS leaves the box Microsoft has no control over it. People install all sorts of programs and make all sorts of "adjustments" to their computers. This makes QA for patches hideously difficult, since MS has to test against such a wide array of third-party apps.
So the argument here is that because the USAF is using an NSA-designed build, they can guarantee a pretty stable environment. MS has a known quantity to test against, which lets them test faster (and presumably better), so they can afford to roll those patches out earlier. They then spend the next few weeks trying to make sure their patches work on Everything Else. One of the hopes cited in the article I read is that this will encourage other entities, like banks and such, to adopt the NSA's build (or at least model their own after it). That will, of course, enable Microsoft to expand its "early release" program, making them more money, but it may also lead to better security across the board. As we all know, a good sysadmin can secure anything, even a Windows box. Well, if you aren't a good sysadmin, maybe you can copy one and get similar effects, right?
That's their line. It does make sense, though I personally would rather see MS release all their patches after minimal QA, then a month (or so) later release "improved" versions. That way, if the patch breaks some third-party program, at least the folks who don't use that program can get the benefits. MS does this sometimes already. Of course, my expectation is that if they did this with every patch, that "month" wait would be closer to two or three months, and often the updated patch would never come out at all.
It could probably be done. You'd need to layer something like FreeBSD's unionfs on top of the running system and back it by memory. When the memory disk gets near full, or the user requests it, it could then be turned into a session and burned to disk, at which point the MD would be emptied and readied for more changes. However this would require remounting the filesystem so the kernel notices the changes you just made, or altering the kernel so it's done automatically. This would be cake for/var or/usr or whatever, harder for/. Remember the days of e2fs when after fsck did something it would say "REBOOT LINUX"? Making changes to a "read-only" filesystem can cause kernel panics.
The easier way of doing it would be with a more robust filesystem like UDF. (Although Linux does support e[23]fs etc. on CDs, you probably can't boot from it.) With enough driver support the CD drive would become just another block device, eliminating the need for hacks like the above. This was, in fact, the case with my SCSI DVD-RAM drive. I just ran mke2fs on it and treated it like a (slow) hard disk. I could even fdisk it.
Note that by allowing this sort of thing you lose some of the benefits of a read-only live CD, in particular its immunity to viruses and such. You'd have to be sure to use CD-R discs, or use the live CD only in a CD-ROM (and not a CD-R or CD-RW) drive.
This may sound really dumb, but isn't it up to the guy who wrote the vulnerability in the first place to fix it?
There is a time gap between when a bug is first discovered and when it is fixed. There is an even bigger gap between when a bug is fixed and when users actually bother to install the patch. Helping to prevent buffer overflows and the like will limit the problems caused by those gaps.
how the hell do you detect an overflow?
Memory is allocated using a library call like malloc(). Debugging tools will trap malloc() and actually allocate slightly more memory than is asked for, then write a signature before and after the buffer. It will then periodically check those signatures to see if they are still there. If they aren't - like because a program overwrote them with its own data - it means there's a buffer overflow. You can also use the CPU's virtualization hardware to spot some kinds of buffer overflows or other errors (like trying to read from a page that was allocated but never written to). There are other methods, but that's the most common and probably the easiest to understand.
I remember when I was younger first hearing about these races and getting excited. It sounded very high-tech and neat, and plus it was a car race - who doesn't like car races? I was, naturally, very disappointed when I learned how slow these things go. My parents explained to me that they couldn't go faster because they didn't have enough energy to do so. At that point I just thought, "Then why are we wasting time on this?"
So I'm not sure these guys are really promoting anything. I strongly suspect their races are having the exact opposite effect, in fact: convincing people that solar technology is nowhere near ready for prime-time. Instead of showcasing stuff solar tech can do that nothing else can, they're showcasing the stuff it does really, really poorly.
As an academic project, I think this is great. I'd love to be involved in it and I'm sure I'd learn a lot just from following it closely. But as PR? Not even close.
No. It can't. We have two concrete examples in this very page - one provided by Wikimedia, one provided by me - which directly contradict your statement. Maybe under some circumstances MySQL can handle reboots, but it's been proven already that it can't always do so. Perhaps your MySQL experience is not with high-load applications (at least not the level of load Wikimedia and my employer see).
BTW, write a damn script. Mysql was written for unix, unix thrives on scripts. If you can't handle writing a script, why the hell are you a DB admin?
Because the process doesn't lend itself well to scripting. For example, MySQL automatically releases locks when you close your connection to the DB. Presumably this is to avoid deadlocks and for other good reasons, but it's not trivial to write a script to do that. Also, since this is an important system, we don't like the idea of trusting computers to handle its repair: we want someone knowledgeable monitoring every step in case something doesn't work exactly right. I can of course sit there and watch the script do its thing, but that defeats the purpose of scripting the process in the first place.
Regardless, the difficulty of the task is not the main issue. The main issue is that we are dealing with north of 1GB of data here, and on busy servers on a busy network that means restarting replication takes an hour or longer. So not only is performance reduced by 33% when we take the slaves offline one at a time, performance is reduced further by the traffic of tar/scp in the background. Not to mention the fact that, because we have a lock on the master's DB, so you can't even consider the DB cluster fully functional.
We have a similar problem at work. There we don't endure database corruption, we just get broken replication. It appears to be working, but it actually isn't. So we have to take the master offline (actually just acquire a write lock on the DB, it can still answer SELECTs), tar up its (massive) database, scp it to the slaves, start the master, stop the slaves, untar the database, restart the slaves, and restart replication. The entire process can take several hours and it's easy to make mistakes. We put stickers on our MySQL servers saying "DO NOT REBOOT WITHOUT CONTACTING OPS MANAGEMENT," though unfortunately faulty DIMMs are illiterate.
I don't know if PostgreSQL has similar problems, but I very much doubt that Oracle or DB2 do. I know that improved failover support has been a target of the PSQL developers for a little while now, so while it may not be on par with Oracle and DB2 it's probably closer than MySQL. At least for now.
I wish this had prompted management to consider alternatives to MySQL, at least for our mission-critical database servers, but unfortunately it hasn't. They don't even see that we could sell an enterprise-level RDBMS as a significant feature - we're a webhosting company - and charge through the nose for it. Oh well. They don't listen to peons like me, they just make me fix MySQL replication every two weeks.
Maybe I'm missing something, but this doesn't seem like anything new. Google for HackerDefender, I'm sure you'll find some relevant links. It intercepts the appropriate system calls to make itself completely invisible: it hides its processes as it's running, it hides the services that start them, etc. I've been seeing it on my employer's Windows servers for quite some time. There are ways to clean it, though they could of course be circumvented as well. The foolproof way to remove it is to boot from a special Windows boot CD and delete the files it uses.
Unless there's something really new and complex going on here, not only is this not new, but IT professionals already have ways of dealing with it. In our case, on a live system with one reboot required. I wouldn't call it minor, certainly (10 minutes of downtime is 10 minutes of downtime), but... hell, if script kiddies have been using this for months and months...
Multiple ways of doing the same thing. (menu, right-click context, shortcut keys, CLI)
But be careful not to overwhelm the user. Basically, make it so that the first thing they try works and then they never even realize there are other options. The worst programs are the ones which sometimes want you to use the menu bar, sometimes use a popup menu, etc.
Don't give the user to much choices at the same time, split into multiple windows if choices get above 3-6 works for me.
The key here is prioritization and logical grouping of choices. The most commonly-accessed options should all be in the same place on the first page. But don't just bury less-used options either, or nobody will know where to find them when they're needed. So after the first layer or two of "common options," try to group stuff logically so that someone who knows what he needs can find it in five seconds at most. If you want an example of how not to build a Preferences dialog, check out Newsleecher.
One related point is how to ask users for data. The wizard approach is good for users who don't always know what they want. It provides a simple way to extract information and make decisions on what to ask next based on previous responses. But it sucks for speed. For a specialized app, like this one, your users probably know exactly what information they need to provide. Let them provide it all at once instead of spread out over 10 dialog pages. Since your users probably won't be entering exactly the same type of data all the time, present them with a simple choice, then give them the huge forms. It's not elegant design (lots of duplicated information) but it's fast as hell. And you can always lay out the forms in the same sort of way, so five tabs takes you to the same field no matter what form you're in, etc.
I mean, don't you hate those Web forms where you answer one or two questions per page? And the damn thing seems to go on for hours? Just give me a monster page with all the fields, I'm smart enough to know which ones are relevant and need to be filled in.
Don't bother users with stuff that can be automated.
But make sure it really can be automated! There's nothing more frustrating than a program which does something you don't want it to with no convenient way to stop it. The key here is to know your audience. A common pitfall for developers is to build the application they want, instead of the application the users want. This plagues the OSS world in particular, for obvious reasons.
Allow the user to configure settings, but provide sane defaults.
Excessive configuration is often a crutch for uncertain design principles. If you know what your app needs to do, and you know how people will be using it, then users shouldn't need to configure anything at all. Fine, let them change the sort order of tables or whatever. But if you find your users need to tweak your program extensively before they're happy with it, you've done something wrong.
This goes back to the "lots of options" point as well. The more options you give your users, the more confused they tend to get and the harder it is for them to find the stuff they need to.
Also note that allowing for customization adversely affects consistency. Even simple customizations can adversely affect the ability of one person to use another's copy of the application. In a place where there are more people than computers that's a serious problem. It's also a problem if you want to offer technical support, training, etc. In a way, let MacOS be your guide: "I don't care if it's the best way, but it's the way you're going to do it." And if you're designing an app which builds, at least to a certain extent, on other apps, you should probably choose to do things the same way the others do. That will be the way your users are used to doing things, so unless you expect to see some enormous improvement in efficiency or ease-of-use, don't rock the boat.
You have a couple options for how to go with this. The hardest requirement you've got is capacity expansion: even many high-end RAID cards won't let you do that, though the feature is becoming more common. If you want to save money, you might want to reconsider that requirement. But you can still "emulate" the feature, though what you get won't be quite as good. Basically what you do is set up a regular, non-expandable array using something like RAID5. Then, when you need more space, you build another (non-expandable) array which is the same size of slightly larger. Because disk is getting cheaper every day this will usually not be a problem: spending the same amount 6 months from now will probably net you 20-50% more capacity. Then you make your new array into a linear LVM volume with one subvolume (the entire new array). You copy over all your old data to the volume, then trash the old array, rebuild it, and add it to the LVM volume. The volume isn't redundant and it isn't striped, but both of your actual arrays are, so you still have pretty good performance and reliability. If this level of complexity doesn't really appeal to you, you can of course just mount both arrays separately and use your brain to fill them up evenly. (Your brain or a bunch of symlinks.)
As far as building the actual arrays goes, that's easier because there are fewer choices. The cheapest (at first blush) approach is to use something like external USB or Firewire disks and software RAID. You don't need an expensive controller card, hot-swap backplane, power supply, or case: the intiial cost of investment is low. The downside is that USB and Firewire disks are much more expensive per GB than regular internal SATA or IDE disks. And while the interface is capable of being faster than SATA/IDE, I suspect those external disk boxes are going to be a good bit slower than regular disks. Still, if you are building an array piecemeal, buying one or two disks at a time, this is the approach I would recommend.
If you have resigned yourself to spending Big Money in one lump purchase, then you will want to do it "right." Buy a hardware RAID card, a hefty power supply, and a case with lots of room. Since cost is a factor, you will want to use SATA or IDE. I recommend SATA simply because cable management is so much easier than with IDE, and the drives are only 5% or so more expensive now; but YMMV, it doesn't really matter which you choose. As for the RAID card to use, decide on what OSes you might be using with the server. If you want FreeBSD, you are pretty much limited to 3ware. If you want Linux, then 3ware and Adaptec should both be fine. Windows is naturally supported by everyone. You'll note that I haven't mentioned the "buy a lot of cheap 2-port PCI controllers and do software RAID" option. Well, not only do those cheap controllers suck ass, many of them only support one or two controllers per machine. You can do it, but it's such a headache I'd rather just drop an extra $200 or so on a many-port RAID card.
For the case to get, the simplest approach is to buy a regular desktop case with a good number of 5.25" bays. You can then buy a "5-to-3 hot-swap backplane," which is a box that fits in three adjacent 5.25" bays and has five hot-swap trays for hard disks. My reason for recommending this isn't really cost, it's that it's tough to find cases which come with hot-swap IDE/SATA bays. SuperMicro sells some, which you can find on Newegg. I have one and it's great. But remember that many of those cases "support dual Xeon processors," which means they come with special 24-pin power supplies and won't work with your P4/Athlon motherboard without an adapter (which is impossible to find). As a final caution, if you are putting in lots of hard disks, make sure you get a case with (a) lots of fans; and (b) a hefty (450W or greater) power supply. I'm using a 450W P/S in my case, because of the above-noted Xeon P/S issue, and it's not sufficient to power four 7200rpm SATA disks: under heavy load one
This is the way most rackmount servers work. Because of space concerns they can't easily fit a regular heatsink/fan combo in the case, especially with 2U (3.5" tall) or 1U (1.75" tall) cases. So they put a bunch of fans on the front or rear (or both) of the chassis and use a plastic duct to route the air over the processors and vent it out of the case. It also seems to be a common tactic among the low-end home desktop systems: the power supply fan is ducted over the CPU and out the back of the case. The motivation there isn't space or cooling efficiency, it's cost (fewer fans) and noise level (again, fewer fans).
I think this is increasingly the way we are going to have to go as we try to squeeze a little more life out of air-cooling. But I really think we are going to have to move to water cooling soon. I'm a little surprised it isn't already happening with servers: due to the space constraints, they would be the biggest beneficiaries. And due to the efficiency benefits, you could cool several servers - maybe even an entire rack - with one radiator (assuming of course that the radiator is designed for that). Smaller installations, with only a few racks, might even be able to eliminate expensive dedicated AC units.
Even in the home market, you could cool a radiator pretty well with a small number of large, low-RPM fans. It should result in a cooler system. My understanding is that at present water cooling is not much quieter than air cooling, if it's quieter at all, but I think that's more a limitation of currently implementations (which target the insane overclocker crowd, who don't care as much about noise).
Well, okay. You're free to define the word however you want. The definition I provided is, however, from dictionary.com. While I did choose the most "open" definition they offered, none of the others differ substantively with respect to this issue. I encourage you to look for yourself, maybe I missed something.
The key meaning to the word, I think, is ownership, or more specifically the legal right to sole ownership. That's pretty much exactly what patents are. The fact that a patent may be licensed royalty-free in perpetuity does not confer ownership of the patent to any third parties, which is the point I was trying to make. Let's say I give you a binding legal promise that you can live in my house for free for as long as you want. Even better, I give you the legal right to let anyone else live in the house too, whether you're there or not. Does this mean you own my house? No: As a matter of law, I still own the house, though in reality my ownership is pretty worthless. In a similar vein, patents which are given away the way IBM has done are still proprietary, because they are owned by IBM. It's just that the terms of their license with you means that they are effectively open.
You could consider it sophistry, which I guess you do, but... One of the benefits to this approach is that you know you aren't infringing on any other patents. Or at least if you are, you have a sure-fire legal defense for doing so. I think the term is "defensive patenting:" you patent something and then give it away for nothing because you're afraid someone else will patent it first and then sell it (or let it languish in some filing cabinet somewhere). This sort of patenting is very much in line with the ideals of both free and open source software. OK, RMS et al. might not like using the system even if it's just to beat the system, but I tend to start from the premise that the patent system ain't going away any time soon.
Proprietary just means "owned by a private individual or corporation under a trademark or patent." Consider IBM's recent "freeing" of a bunch of their patents for use in open source software. That would be proprietary technology in an open source product. It's not a contradiction at all.
I also think that when you say "Open" what you actually mean is something closer to "Free." Open Source is a notoriously pragmatic term, whereas Free Software aims more for philosophical freedom ("free as in speech"). Proprietary stuff is far less likely to be Free than it is to be Open.
I think I just confused myself with that one; maybe I should put away this RMS-autographed Ouija board I've got on my desk.
But the net does not suffer bullshit or propaganda gladly, and instead skewers it on innumerable web forums (you're soaking in it now!).
On the contrary. The Internet doesn't discriminate against anything. I can post the most nonsensical, baldfaced lie in the world, and the Internet gives it the same weight as everything else.
It really amazes me that you can use the Internet enough to discover Slashdot and not know that the overwhelming majority of the Internet is the babbling of depraved lunatics. I definitely include Slashdot in that assessment, because naturally some of the moderators are themselves depraved lunatics. (Or, more generously, they are just succumbing to the herd instinct and modding up anything which seems like the Slashdot Party Line.)
Once broadband becomes cheap, look for America to become more like Sweden or Denmark--more united against the rich and powerful....
As someone who hopes to become both rich and powerful someday, I hope that doesn't happen. Fortunately, as most Americans at least share my ambitions, I doubt it will happen in my lifetime.
If the result of this is MS fully opening the MS Office file formats, so that every other office suite out there can read and write them with 100% compatibility, then that's great! It's not as good for open source as mandating the use of e.g. OO.o would be, but it's still good, and more importantly it focuses more on freedom. (I don't see how being forced to use OO.o would be any better than being forced to use MS Office. I still don't have a choice either way.)
I did RTFA and it's a little unclear as to whether this is what's actually happening or not, but I can certainly hope.
Slashdot Mirror
It could be useful for other things. For example, doing tape backups over the network. A full backup of a 50GB drive over the network is going to take two hours (or longer). During that time your server's hard disk is going to be thrashing like crazy and performance will be shot to hell. But if you've got two disks, then you just "break" the array (making one of the disks readonly) and backup from just one disk. Minimal performance hit, you get the benefits of RAID1 (and snapshots), and it's a pretty simple system so there's not too much that can go wrong. We have a couple hundred servers at work that use this approach and it's amazingly convenient and problem-free. A lot less flaky than most RAID hardware too, sad to say. (I'm looking in your direction, Mylex and LSI.)
It's possible to control every aspect of a Windows computer (except MMC, dammit) from the keyboard. Except for third-party apps, which are often broken - but you can't blame MS for that. Most keyboard shortcuts have been Windows-standard since the 3.0 days, it's not MS's fault that so many developers are too lazy to implement them.
I would like visual feedback a la MacOS (or Firefox, for that matter) for typeahead find in folders though. But oh well, I type fast enough for it not to be a problem.
Also, are the clipboard shortcuts really different? I seem to remember they're Option+ZXCV, which is not terribly different from Ctrl+ZXCV. It's some vendor-agnostic usability standard; I remember IBM making a big deal of it back in the OS/2 days, when Windows (3.0) wanted you to use Ctrl/Shift+Insert and Shift+Delete for copy, paste, and cut, respectively. (They still work, try them out.)
Incidentally:
It's hard to be both most important and second-most important at the same time. Yet this is apparently a feat both AMD and Intel have managed. I guess this is a byproduct of their research into quantum computers. ("Alright! We're on top! Oh shit, I just changed our importance by measuring it.")Or MS can say "The fines for noncompliance will be lower than the cost of damage to our business if we do comply. So we're going to eat the fines." Even if that were true, eventually the EU would either crank up the fines or simply bar MS from selling Windows as-is, in which case we're back to option 1.
Say I claim that the sky is red and offer evidence to that effect. If my evidence is inadequate, then it's not proof. If my evidence proves that the sky is definitely not red, then it's contra-proof: it proves the opposite of what I am claiming.
It would be less awkward to say "proof negative," but the contra- prefix is common in philosophical circles where this sort of fine distinction is usually relevant.
I would expect that as these satellite-repairing bots become more advanced we will see newer satellites which are designed to allow easier repairs.
I love how asserting any paranoid theory on Slashdot gets you modded insightful.On an unrelated topic, there is a herd of purple dancing elephants who live in my closet, but light makes them invisible. Sometimes at night, in the dark, I can see them laughing at me. My friends say it's just the outline of my clothes being blown by the air conditioner, but I know they're lying. It just proves that my friends are all on the payroll of the elephants. We're fast approaching a world run by a shadowy heliotrope proboscidean ecdysiast cabal operating out of my bedroom. It may already be too late to stop it.
These sorts of problems are very common in the Northeast where the states are so small. On the one hand, the ruling seems almost fair: you're getting the benefits of NY (mass transit, cops, etc) but not paying a dime in taxes. But on the other hand, any taxes an employer pays are effectively taxes levied on all its employees (and customers!), so I'm not sure that really holds water. (This is that "double taxation" that some flat-tax advocates talk about.)
Anyway, that would let open source developers use code contributed by IBM and other big IP holders free of worry about being sued some day. Even if IBM (for example) contributed code using patents it doesn't own (such as SCO), developers would have an extremely good defense. "IBM and its horde of patent lawyers told me I could use it." Continuing with my hypothetical situation, it would also apply to people who use the software, so long as they have agreed to the GPL, and so it would protect Linux users and not just developers.
That's my thinking. But neither I nor anyone else who has posted to this story has any clue what will really be in GPLv3, so... you know, don't be surprised if I turn out to be totally wrong.
The former is important because you will need to unplug those bastards occasionally, and I have some Radio Shack cables that I have to disconnect with a wrench and significant amounts of muscle. I'm concerned about literally breaking the connector from the amount of force I have to apply. From the other direction, if they come off too easily you will have to tape them to the TV or something, which doesn't impair signal quality but is annoying.
The latter is important not because of interference or anything, but because stuff will happen to the cables. You'll vacuum it, your pets will chew on it, you'll get crap on it, etc. It's a pain in the ass to get behind your TV to replace frayed or broken cables, so you should try to minimize it.
Note that neither of these considerations imply that you should buy Monster cables, just that you should avoid the extremely inexpensive (read: cheap) ones. So instead of buying the $30 Monster cables or the $5 generic cables, buy the $15 name-brand-but-not-Monster cables. Belkin seems like a pretty good target for that, and you could in fact do better if you find a place that lets you return cables. (You obviously have to open up the package to try them out, and some places don't want to take them back after you do that.)
I do always laugh when I see the rusted-out '85 Honda Civic with the brand new shiny 20" rims and chrome exhaust tips. "You just spent $2500 on mods, and a Civic is the best you can do?"
Since we know the pain of being forced to use an OS we dislike, let's try to be the Better Man here. If the tables someday turn, let's continue to encourage that apps get ported to every OS under the sun. Not because it helps us (it wouldn't, in this hypothetical world), but because it's the right thing to do.
By the same token, I support the PSQL folks porting their excellent software to Windows. Let those who, for whatever reason, want to run Windows make that choice themselves and still be able to run PSQL.
So the argument here is that because the USAF is using an NSA-designed build, they can guarantee a pretty stable environment. MS has a known quantity to test against, which lets them test faster (and presumably better), so they can afford to roll those patches out earlier. They then spend the next few weeks trying to make sure their patches work on Everything Else. One of the hopes cited in the article I read is that this will encourage other entities, like banks and such, to adopt the NSA's build (or at least model their own after it). That will, of course, enable Microsoft to expand its "early release" program, making them more money, but it may also lead to better security across the board. As we all know, a good sysadmin can secure anything, even a Windows box. Well, if you aren't a good sysadmin, maybe you can copy one and get similar effects, right?
That's their line. It does make sense, though I personally would rather see MS release all their patches after minimal QA, then a month (or so) later release "improved" versions. That way, if the patch breaks some third-party program, at least the folks who don't use that program can get the benefits. MS does this sometimes already. Of course, my expectation is that if they did this with every patch, that "month" wait would be closer to two or three months, and often the updated patch would never come out at all.
The easier way of doing it would be with a more robust filesystem like UDF. (Although Linux does support e[23]fs etc. on CDs, you probably can't boot from it.) With enough driver support the CD drive would become just another block device, eliminating the need for hacks like the above. This was, in fact, the case with my SCSI DVD-RAM drive. I just ran mke2fs on it and treated it like a (slow) hard disk. I could even fdisk it.
Note that by allowing this sort of thing you lose some of the benefits of a read-only live CD, in particular its immunity to viruses and such. You'd have to be sure to use CD-R discs, or use the live CD only in a CD-ROM (and not a CD-R or CD-RW) drive.
So I'm not sure these guys are really promoting anything. I strongly suspect their races are having the exact opposite effect, in fact: convincing people that solar technology is nowhere near ready for prime-time. Instead of showcasing stuff solar tech can do that nothing else can, they're showcasing the stuff it does really, really poorly.
As an academic project, I think this is great. I'd love to be involved in it and I'm sure I'd learn a lot just from following it closely. But as PR? Not even close.
Regardless, the difficulty of the task is not the main issue. The main issue is that we are dealing with north of 1GB of data here, and on busy servers on a busy network that means restarting replication takes an hour or longer. So not only is performance reduced by 33% when we take the slaves offline one at a time, performance is reduced further by the traffic of tar/scp in the background. Not to mention the fact that, because we have a lock on the master's DB, so you can't even consider the DB cluster fully functional.
I don't know if PostgreSQL has similar problems, but I very much doubt that Oracle or DB2 do. I know that improved failover support has been a target of the PSQL developers for a little while now, so while it may not be on par with Oracle and DB2 it's probably closer than MySQL. At least for now.
I wish this had prompted management to consider alternatives to MySQL, at least for our mission-critical database servers, but unfortunately it hasn't. They don't even see that we could sell an enterprise-level RDBMS as a significant feature - we're a webhosting company - and charge through the nose for it. Oh well. They don't listen to peons like me, they just make me fix MySQL replication every two weeks.
Unless there's something really new and complex going on here, not only is this not new, but IT professionals already have ways of dealing with it. In our case, on a live system with one reboot required. I wouldn't call it minor, certainly (10 minutes of downtime is 10 minutes of downtime), but... hell, if script kiddies have been using this for months and months...
One related point is how to ask users for data. The wizard approach is good for users who don't always know what they want. It provides a simple way to extract information and make decisions on what to ask next based on previous responses. But it sucks for speed. For a specialized app, like this one, your users probably know exactly what information they need to provide. Let them provide it all at once instead of spread out over 10 dialog pages. Since your users probably won't be entering exactly the same type of data all the time, present them with a simple choice, then give them the huge forms. It's not elegant design (lots of duplicated information) but it's fast as hell. And you can always lay out the forms in the same sort of way, so five tabs takes you to the same field no matter what form you're in, etc.
I mean, don't you hate those Web forms where you answer one or two questions per page? And the damn thing seems to go on for hours? Just give me a monster page with all the fields, I'm smart enough to know which ones are relevant and need to be filled in.
But make sure it really can be automated! There's nothing more frustrating than a program which does something you don't want it to with no convenient way to stop it. The key here is to know your audience. A common pitfall for developers is to build the application they want, instead of the application the users want. This plagues the OSS world in particular, for obvious reasons. Excessive configuration is often a crutch for uncertain design principles. If you know what your app needs to do, and you know how people will be using it, then users shouldn't need to configure anything at all. Fine, let them change the sort order of tables or whatever. But if you find your users need to tweak your program extensively before they're happy with it, you've done something wrong.This goes back to the "lots of options" point as well. The more options you give your users, the more confused they tend to get and the harder it is for them to find the stuff they need to.
Also note that allowing for customization adversely affects consistency. Even simple customizations can adversely affect the ability of one person to use another's copy of the application. In a place where there are more people than computers that's a serious problem. It's also a problem if you want to offer technical support, training, etc. In a way, let MacOS be your guide: "I don't care if it's the best way, but it's the way you're going to do it." And if you're designing an app which builds, at least to a certain extent, on other apps, you should probably choose to do things the same way the others do. That will be the way your users are used to doing things, so unless you expect to see some enormous improvement in efficiency or ease-of-use, don't rock the boat.
As far as building the actual arrays goes, that's easier because there are fewer choices. The cheapest (at first blush) approach is to use something like external USB or Firewire disks and software RAID. You don't need an expensive controller card, hot-swap backplane, power supply, or case: the intiial cost of investment is low. The downside is that USB and Firewire disks are much more expensive per GB than regular internal SATA or IDE disks. And while the interface is capable of being faster than SATA/IDE, I suspect those external disk boxes are going to be a good bit slower than regular disks. Still, if you are building an array piecemeal, buying one or two disks at a time, this is the approach I would recommend.
If you have resigned yourself to spending Big Money in one lump purchase, then you will want to do it "right." Buy a hardware RAID card, a hefty power supply, and a case with lots of room. Since cost is a factor, you will want to use SATA or IDE. I recommend SATA simply because cable management is so much easier than with IDE, and the drives are only 5% or so more expensive now; but YMMV, it doesn't really matter which you choose. As for the RAID card to use, decide on what OSes you might be using with the server. If you want FreeBSD, you are pretty much limited to 3ware. If you want Linux, then 3ware and Adaptec should both be fine. Windows is naturally supported by everyone. You'll note that I haven't mentioned the "buy a lot of cheap 2-port PCI controllers and do software RAID" option. Well, not only do those cheap controllers suck ass, many of them only support one or two controllers per machine. You can do it, but it's such a headache I'd rather just drop an extra $200 or so on a many-port RAID card.
For the case to get, the simplest approach is to buy a regular desktop case with a good number of 5.25" bays. You can then buy a "5-to-3 hot-swap backplane," which is a box that fits in three adjacent 5.25" bays and has five hot-swap trays for hard disks. My reason for recommending this isn't really cost, it's that it's tough to find cases which come with hot-swap IDE/SATA bays. SuperMicro sells some, which you can find on Newegg. I have one and it's great. But remember that many of those cases "support dual Xeon processors," which means they come with special 24-pin power supplies and won't work with your P4/Athlon motherboard without an adapter (which is impossible to find). As a final caution, if you are putting in lots of hard disks, make sure you get a case with (a) lots of fans; and (b) a hefty (450W or greater) power supply. I'm using a 450W P/S in my case, because of the above-noted Xeon P/S issue, and it's not sufficient to power four 7200rpm SATA disks: under heavy load one
I think this is increasingly the way we are going to have to go as we try to squeeze a little more life out of air-cooling. But I really think we are going to have to move to water cooling soon. I'm a little surprised it isn't already happening with servers: due to the space constraints, they would be the biggest beneficiaries. And due to the efficiency benefits, you could cool several servers - maybe even an entire rack - with one radiator (assuming of course that the radiator is designed for that). Smaller installations, with only a few racks, might even be able to eliminate expensive dedicated AC units.
Even in the home market, you could cool a radiator pretty well with a small number of large, low-RPM fans. It should result in a cooler system. My understanding is that at present water cooling is not much quieter than air cooling, if it's quieter at all, but I think that's more a limitation of currently implementations (which target the insane overclocker crowd, who don't care as much about noise).
Let me come with you, Thylvethter! I might be of thome athithtanth if there ith a thudden crithith!
The key meaning to the word, I think, is ownership, or more specifically the legal right to sole ownership. That's pretty much exactly what patents are. The fact that a patent may be licensed royalty-free in perpetuity does not confer ownership of the patent to any third parties, which is the point I was trying to make. Let's say I give you a binding legal promise that you can live in my house for free for as long as you want. Even better, I give you the legal right to let anyone else live in the house too, whether you're there or not. Does this mean you own my house? No: As a matter of law, I still own the house, though in reality my ownership is pretty worthless. In a similar vein, patents which are given away the way IBM has done are still proprietary, because they are owned by IBM. It's just that the terms of their license with you means that they are effectively open.
You could consider it sophistry, which I guess you do, but... One of the benefits to this approach is that you know you aren't infringing on any other patents. Or at least if you are, you have a sure-fire legal defense for doing so. I think the term is "defensive patenting:" you patent something and then give it away for nothing because you're afraid someone else will patent it first and then sell it (or let it languish in some filing cabinet somewhere). This sort of patenting is very much in line with the ideals of both free and open source software. OK, RMS et al. might not like using the system even if it's just to beat the system, but I tend to start from the premise that the patent system ain't going away any time soon.
I also think that when you say "Open" what you actually mean is something closer to "Free." Open Source is a notoriously pragmatic term, whereas Free Software aims more for philosophical freedom ("free as in speech"). Proprietary stuff is far less likely to be Free than it is to be Open.
I think I just confused myself with that one; maybe I should put away this RMS-autographed Ouija board I've got on my desk.
It really amazes me that you can use the Internet enough to discover Slashdot and not know that the overwhelming majority of the Internet is the babbling of depraved lunatics. I definitely include Slashdot in that assessment, because naturally some of the moderators are themselves depraved lunatics. (Or, more generously, they are just succumbing to the herd instinct and modding up anything which seems like the Slashdot Party Line.)
As someone who hopes to become both rich and powerful someday, I hope that doesn't happen. Fortunately, as most Americans at least share my ambitions, I doubt it will happen in my lifetime.I did RTFA and it's a little unclear as to whether this is what's actually happening or not, but I can certainly hope.