M$ is all about their Windows/Office monopoly. On the PC. Period. For almost two decades now, the company's whole business strategy has really just been about maintaining that monopoly. It's what they've always done best.
Sure, they dabble in many other areas, but unless it soon starts to look like it has the potential to make them billions, just like Windows and Office, they quickly lose interest.
Innovation? Nope. They just buy other companies that look interesting, assimilate the knowledge, achieve relatively little as a result and then move on. Seriously, I can't think of anything unique that they themselves have ever come up with besides M$ Bob, which was a total flop, and perhaps a miserable animated paperclip. All of their other ideas have either been bought or stolen.
Seen in this light it's easy to predict when M$ will die: when the PC dies. Once their traditional monopoly has faded, M$ will be just another software company struggling with the competition... except that they never learned how to innovate -- they've never had to!! Oh, of course there are more people at M$ who know that and fear the inevitable, but they can't change anything. Their ship's course was fixed long ago and it can never be altered. Ray should have known that and avoided them like the plague, but I suppose he was just being stupid or greedy... or both. Within that company, however, he's hardly a unique case; just more well known.
My servers have become ever more complicated over the years: first Bind9, ISC-DHCP, NFS, Squid, Exim4, Spamassassin, ClamAV, CUPS, Apache2, PostgreSQL, Samba, Wide-DHCPv6, NTP/GPSD, and Asterisk to name a few, but now I've thrown out NFS and added Kerberos, OpenLDAP and OpenAFS (the fabulous Andrew File System). However, even though my Debian GNU/Linux server can easily handle this entire load, there are drawbacks to running all of these processes on a single physical server machine. The ones I see involve security and compatibility.
From a security standpoint, the problem is that if the wrong people ever manage to gain root access through some weakness in, say, Apache2 or Exim4, then they've got access to everything. As for compatibility, whenever the decision is made to upgrade the OS of a system that has only one OS, many if not all of the applications it runs also have to be upgraded. If any major problems are encountered, the ability to run certain applications may be lost, or it may even be necessary to turn back and reinstall the old OS, which may represent a terrible waste of time, money and effort. This might also be referred to as a continuity problem.
To address these issues, I've started working with KVM and Linux-VServer: the former is like a more traditional virtual machine, always requiring set chunks of memory and disk space for its VMs, while the second is like working with chroot jails on steroids. Despite all of their magic, however, both were easier to configure and maintain that I expected. So, as long as there's enough memory, it's now possible to set up server machines that start out running little more than KVM and libvirt-bin, along with the usual LVM tools for disk management. These physical machines can then be used to launch a number of KVM virtual machines, each of which can in turn run one or more Linux-VServer machines.
In the above scenario, it's possible to isolate many different applications -- especially those that are relatively vulnerable or need to be kept extra secure (e.g. Kerberos). If the bad guys do gain remote access to any of the more vulnerable applications, then it's more likely that the damage they do will be limited. In addition, upgrading becomes much less of a risk, since the virtual systems can in principle be upgraded individually; if one application does not upgrade properly, it doesn't have to hold the rest back. Even better, if you always keep a gigabyte or so of memory in reserve along with a fair amount of disk space, then you can always opt to replace older virtual systems as opposed to upgrading them; a good way not to burn your bridges behind you in case things go wrong. Finally, working this way means complicated upgrades will no longer have to be done all at once: they can be spread out over days, weeks or months. The only drawback is that you're left with multiple operating systems to maintain instead of just one per physical machine. But, that now seem like a small price to pay.
As for virtualization on the desktop, I've been using VirtualBox for the past so many years, but I have no idea how long that's going to last, because I don't think we can trust Oracle. I guess I'll end up using KVM and Linux-VServer for that as well. I suspect that it will be better to do that now while we have the choice, as opposed to later when we have none.
* Once every day at 10 hours, 10 minutes and 10 seconds. But okay, that moment only lasts a second.
* On the 10th day of every month at 10 hours and 10 minutes, lasting sixty seconds.
* On the 10th hour of the 10th day of October every year, lasting one hour.
Today's date is only special because the 10/10/10-moment lasts an entire 24 hours, which is way more time than most people will care to dwell upon in its specialness. True, it only occurs once a century, but just remember that it can't hold a candle to October of the year 1010, which lasted an entire month and won't be seen again for another 9 millenia.
There are, of course, plenty more 10/10/10-moments for nerds to look forward to in the far, far future that will last a lot longer than that (a year, a century, a millennium...), but then we would have to assume that the creatures of this planet will still be hanging on to the Gregorian calendar after all that time, which is perhaps doubtful.
Perhaps weeds would be able to help us out here, since they grow quickly and don't seem to require any fertilizer. Needing only water, farmers could be subsidized to plant endless crops of some particularly fast-growing (genetically tailored?) varieties that would subsequently be harvested and buried. Perhaps burial would also help to prevent these oxygen-deprived organic masses from turning into sources of methane, which as a greenhouse gas is 25 times more potent than CO2. Or, maybe the methane could be trapped and burned to produce energy. This would produce CO2, but not be nearly as bad as letting the methane escape into the atmosphere.
Everybody knows that before you can shutdown the OS, you first have to start it up! But, as it turns out, that's a mighty complicated process. You first have to ask yourself things like:
Do I have an OS to begin with?
Do I have a computer to run it on?
Do I have an electrical socket to plug the computer into?
Do I have a keyboard?
Do I have a monitor?
Do I have a mouse?
Do I really want to start up my OS?
Does my mother/boss agree that I should be allowed to start up my OS?
Do I have access to a qualified help desk service to help me start up my OS?
Do I have the necessary license to start up my OS?
Do I know how to suck eggs?
Now imagine somebody else asking you these questions... you see how complicated this can become? Luckily, it doesn't have to be that way. All you have to do is license Slashdot's patented OS Startup Procedure and you'll no longer have to lie awake, agonizing about how to deal with all this complexity!
That's what I think my problem is. Recently, a very good friend of mine told me that if I installed OpenAFS on the systems of anyone who is dependent on me for support, he would not be able to help me with that if things go wrong when I'm not around (no, he has no interest in learning about it). Not that I was counting on his support, but that's still frustrating to hear. I worked very hard on getting OpenAFS to work, but I have to admit that, even within the Linux community, it's a lot more complex and far less popular than NFS.
Originally, I was concerned mainly with scalability and security. NFS gets very low marks in these areas, while OpenAFS scores very highly. However, I think my good friend's view has something in common with the position that many large organizations take in these matters: they just want to keep things as average and simple as possible so that they will always be less likely to suffer problems when it becomes necessary to hire or fire IT personnel.
This is frustrating for me. My IT career started when I was 30, it peaked financially seven years later, but nine years after that my technical knowledge is still growing. Hey, I love what I know now! Technically speaking, I feel like I can run circles around my former self. The problem is, most employers aren't interested in what I know now; only whether I'm up to speed on the latest software from M$. There's a difference here: where I grew impatient and decided to solve the basic problems myself, most employers apparently still prefer to wait for M$ to solve them instead.
It's ironic. One of the things that attracted me to IT in the first place was that it allowed for creativity. That's still true, but if you want a career as a system administrator, it looks like you don't want to get too creative.
Win95 seemed promising at first, but then...
on
Windows 95 Turns 15
·
· Score: 4, Insightful
CPU and memory requirements were so much higher, that you basically needed a brand new machine to run it.
Because of the registry, it was no longer possible to copy a program to another machine by simply copying a particular directory structure and a few.ini files. For M$, of course, this was the entire point. Unfortunately...
Because the registry was so easily corruptible, people who used it would regularly see their machine's performance drop and/or encounter regular lock-ups and blue screens, and subsequently find themselves spending hours reinstalling everything. It was no longer possible to fix things by modifying a few.ini files with a text editor.
Because of the registry, which would quickly grow beyond the size of a 1.44 MB floppy disk, the only real backups possible were disk-image backups.
Because the registry could so easily be exploited, the number of species of computer viruses exploded. Without that, the virus industry would certainly not be as successful as it is today.
Because of the registry, it could become next to impossible to get certain complicated machines, particularly fancy laptops, to work properly after installing all of the necessary drivers.
It would not lend itself to the simple remote boot method that was previously so popular with Win311 (well, I do know of one NetWare shop that actually managed this feat anyway, but it was very complex). For many of us who thought we had things licked, this made network maintenance an order of magnitude more complicated.
At the time may career as a NetWare sysadmin was just taking off, so it was another six years before I made the switch to Linux, but for me Win95 marked the beginning of the end of my belief in proprietary software.
What you're worried about is having to deal with IPv6 addresses that are the product of stateless autoconfiguration. Those look daunting for sure, but luckily you don't have to use them if you don't want to (let alone remember them). Specifically, you can can also choose to configure your hosts with fixed IPv6 addresses, either manually, or (much better) with DHCPv6 stateful autoconfiguration. Then you can have IPv6 addresses that are about as easy to remember as with IPv4, except that you'll have more than you'll ever need and they can all be public addresses.
Once they figure out that there's money to be made from jiggering the Internet traffic flow through their networks, there won't be anything to keep all of the ISPs from doing the same thing.
The problem is that as concepts go, Net Neutrality is pretty abstract: if some of us even have colleagues who don't understand what's at stake, we can be sure that the overwhelming majority of ISP customers don't know or care. So, if it's not enforced by law, to expect any individual ISP to voluntarily treat all of its Internet traffic the same would be similar to expecting there to be some large theaters that would be be willing to play movies without showing advertisements. Of course, all those theaters will tell you that if they did that, their prices would have to go up, and that's true. However, the fact is that they never give us that choice, because A) they know that most movie goers don't care anyway and B) they know that the advertisers would not like the viewers to be given that choice -- better to keep things simple!
Yes, really small theaters often don't bother with ads, but that's because they don't sell enough tickets. Advertisers are only willing to pay theaters significant amounts of money if they can be convinced that the ticket sales are high enough. Below a certain threshold there's not enough money in showing ads, so theater owners will often try to increase their ticket sales by advertising that they don't show any advertisements.
In the same way, only small ISPs would advertise Net Neutrality because A) they aren't big enough to convince any significant content providers to make deals with them and B) they can't afford the necessary equipment anyway. On the other hand, in this case there's nothing to prevent a small ISP's upstream service provider from jiggering the traffic. And for that matter, if Net Neutrality were not required by law, where would the ever jiggering stop for sure?
In mid June I set up my latest server based on Squeeze with the expectation that it would go stable this summer. For a while I thought perhaps I had jumped the gun and would be stuck with a relatively unstable system for a longer period, but I guess not.
In particular, I'm happy with Squeeze because I could use it to get my Kerberos-OpenLDAP-OpenAFS system working on both the file server and workstations. Not that I've ever use any FOSS other than Debian for my server, but after my attempts failed to get the latest Ubuntu client to run the necessary client software for this (unfortunately) uncommon, but very capable distributed file system, I suspected the same Debian version for the workstation represented my best chance of success. And sure enough: it worked straight away! Ubuntu may have certain benefits, but it seems that if you want a desktop system that is a little out of the ordinary, Debian is still your best bet.
For a year I used Kubuntu, as well as the Linux Mint version based on it, for my workstation. I switched because KDE on Debian, of which I'm a big fan, had become so unstable in the spring of 2009 as KDE4 was being introduced. I'm grateful to Mr. Shuttleworth that I had this option. I was forced to move back to Debian because there are currently too many bugs in the Ubuntu packages that I need to support a distributed file system based on Kerberos, OpenLDAP and OpenAFS. This all works with Debian (lenny or squeeze), so I figure Ubuntu is just too focused on the desktop to care about it.
That's a pity for two reasons. First, I definitely had it easy for a while as far as the desktop is concerned. I've been back with Debian for a month now and there are still a number of rough edges to my desktop experience: I've spent far too much time adding missing functionality and trying to get it all to behave properly. It's such a waste of effort when you know that it doesn't have to be like that anymore. Linux Mint is so easy, even a relative noob can install it and have all kinds of basic desktop functionality running and configured in just a few hours.
The second reason is because Linux workstations deserve better file server support than just NFS and SMB/CIFS. Imagine an office building that will soon house 2.000 employees and being offered the opportunity to set it up with workstations and servers using only open source software. Would you feel comfortable doing that with NFS or Samba? I wouldn't. OpenAFS, on the other hand -- now that's a capable file system. I know that I would be able to rely on Debian and OpenAFS for the file servers, but I would also prefer a distro for the workstations that would likely result in the lowest number of help desk calls. I doubt that would be Debian, but it would be great if it could be something based on Debian. With OpenAFS and distros like Ubuntu, I figure we're almost there.
From this perspective, I find it really strange that so many long-time Debian users can be so hostile towards Ubuntu. It's not like anyone is forcing them to use it. IMHO, if it's so easy to use that it not only gives normal users the necessary confidence to make the switch from Windows, but also to fix (most of) their own problems afterwards, how can that be a bad thing? Furthermore, if the Ubuntu project continues to succeed where the Debian project has not, perhaps the latter should look to the former for a little inspiration every once in a while.
When my Sun VirtualBox application asked me to upgrade recently to version 3.2.6, I thought nothing of it; sure, always good to stay up to date, right? It proceeded without a hitch, although the only difference seemed to be that the branding had been changed from "Sun Microsystems Inc." to "Oracle" along with some corporate artwork. Was that all? Apparently not. Afterwards, several Linux distros refused to install as virtual machines: somewhere half-way, the process would report a generic failure. Strange, because I had installed the same distros many times before (at least the older VMs I had of them were still working). So, suspecting a bug, I downgraded back to v3.1.8 r61349 and the problem vanished. Pretty weird for what's ostensibly just a cosmetic change. At any rate, I feel this does not bode well for the future of VirtualBox and I've already been looking into the alternatives; perhaps something a little less dependent on some corporate overlord will be a safer choice.
Too many members of the general public are ignorant of science, what its basic tenants are, how it works, why it has been so successful and therefore why it deserves everyone's respect and attention, especially when scientists warn us about things like tornados, the AIDS virus, volcanic eruptions, tidal waves, lead based paint, and releasing too much greenhouse gas into the atmosphere.
Unfortunately, teaching people the facts about the universe we live in is difficult and expensive. But when society fails to educate its members sufficiently about science, to teach them to think critically, then the purveyors of disinformation -- typically organized religions and corporate marketing departments -- are always there to enlighten them with their own versions of the truth.
What can we do about this? First, never cut back on education. An enlightened society is an educated one and maintaining it as such is a endless task. Second, make education accessible to everyone at no cost. Three, we have to be hard on ourselves to ensure that our teachers and educational institutions continue to live up to the highest standards. Four... spend money on marketing facts that are both generally accepted by the scientific community and important to society.
How do we pay for all that? Higher basic taxes, I guess (it will eventually pay for itself), but perhaps also by levying a tax on top of what those purveyors of disinformation spend on advertising.
So much environmental damage has already been done and we're all still strongly addicted to fossil fuels. We're like a bunch of heroin addicts from whom going cold-turkey in basically unthinkable.
When comparing energy sources, what we already have is always cheaper, so why switch? And even if democracy were suspended in the USA, if other countries did not follow suit, Americans would soon have to admit that suddenly going green will have put them at a serious economic disadvantage.
There are no easy fixes. Millions may now be aware that we are collectively flirting with environmental disaster, but that makes no difference if billions don't care. What's needed is for the vast majority of the Earth's inhabitants to agree to give the climatologists the benefit of the doubt and to live on less for quite a while -- voluntarily -- in order to get things fixed.
Unfortunately, we live in a world in which most people strive to accumulate more wealth, but it's concentrated with 1% of the population (who own 95% of everything and are loath to part with it) while most of the rest live in desperate poverty and are only concerned with their own short-term survival (which often includes destroying their environment). Sometimes we cooperate, but mostly we just compete with each another.
I keep thinking that on the largest scale, we're no better than bacteria. I imagine a hypothetical colony in a petri dish, together with some algae: if they controlled their growth so as to strike a balance with the algae, they could potentially live indefinitely in their little closed ecosystem (assuming there was always enough sunlight). However, that would mean that the bacteria would regularly have to force themselves to go hungry and die... seemingly for nothing, because actually there would always be food around. It's much easier for a population to allow itself to grow unchecked, so in reality the bacteria would simply consume all of the algae as fast as possible after which their entire population would die off. Of course, the bacteria could form spores with the last of their strength, but that's not really an option for us.
No, I seriously doubt that humanity is going to fix this problem in an organized, rational and peaceful manner. The alternative? Nature will correct the imbalance for us. There will be war and famine for the vast majority, billions will die, but about 1% of the population will continue to live in relative happiness. Nothing new, really.
For a moment, I thought the same, but then I remembered reading somewhere that insects can only see ultraviolet the way they do because the light-sensitive cells of their compound eyes are so close to the surface. In the case of the mammalian eye, however, evolution has caused the lens to become opaque to most frequencies of ultraviolet light to avoid damage to the retina. So, even if the retina was sensitive more of the UV range, it would still not be able to detect it.
As for the possibility of seeing infrared light, if it were ever to become an option I'm not so sure that I would want to see too much of that range either. For example, it might be interesting to be able to detect objects and other living things by differences in temperature the way a pit viper does, but that would also have serious drawbacks. For example, a hot frying pan might become too painful too look at, and you'd also have to get used to falling asleep with the lights on, so to speak: after closing your eyes, you'd still see the light (the heat) from the inside of your own eyelids!
It's more likely though, IMO, that we will never be able to see infrared directly with our eyes. Snakes, for example, have evolved several different methods of detecting warm-blooded prey (parallel evolution), but those solutions all involve heat-sensitive patches of skin or scales (on, or between the labial scales of many boas and pythons, and of course the famous pit organ) -- it never involved the eyes. Other animals that see better than we do at night make use of a layer behind the retina, called the tapetum lucidum, that reflects light that has managed to pass undetected through the retina back out again, thus increasing the chances of it being detected on the second pass. For those who would be willing to pay for it, this might be a more interesting feature to add to the human eye, but it still would not help to detect infrared light.
Right now the situation with health insurance in the USA is so awful that any change is welcome, just as long as that change does not actually come from the health insurance industry itself.
The problem with the health insurance industry in the USA is that it is hardly regulated at all. It's unforgivable that the law currently allows them to choose not to insure people based on their personal medical histories, and even worse that people who are insured (and pay the highest premiums of any developed nation) can be (and often are) given such a hard time when health insurance companies decide that they simply don't want to pay.
Still, so many people in the States seem reluctant to accept any new government proposals for change, because they fear that their already bad situation may actually get worse. They shouldn't be. The only ones who really have anything to fear are the health insurance companies, who are, incidentally, fighting the proposed legislation tooth and nail. That alone should allay people's fears, but amazingly, few seem to notice.
This coming weekend, if Obama fails to pass his health care plan, it will be a very sad day indeed for the American people. But, I won't blame Obama: as president, he can only do so much when congress refuses to cooperate. I want to see that Americans are still capable of helping themselves, but I fear they are not.
Let's face it, America has been a plutocracy for decades: it's run by the rich top percentile who cynically manipulate public opinion to their advantage... not hard to do in a country where most people have a pretty miserable education. Wake up, Americans: your country's star is fading fast.
Sorry, but I neglected to mention that the SSD in question was indeed both an early and a cheap model. I believe the manufacturer was OCZ, but I forget which model it was. This may subtract some weight from my original assertion, but I don't think all of it.
By the end of 2008 the Intel X25-M was supposed to be the best thing around, but even that model suffered from a form of low-level fragmentation that was the result of using both wear leveling and write combining. These are workarounds for problems that are inherent to SSDs that use NAND flash chips -- especially of the MLC variety -- and a lot of the competition involves trying to develop controller chips that are better at keeping that inevitable fragmentation under control. Only, I don't think any of them will ever find a practical way to really solve the issue once and for all. So, I'm getting tired of hearing about the "latest storage breakthroughs" that are based on NAND flash technology.
Okay, then why don't the chip manufacturers just ditch NAND all together and switch to a different technology with better prospects? Well, first because there's no demand for anything else (um, they don't want to invest the money that will create the demand), and second for the same reason the hard disk manufacturers don't want to quit: they first want to get as much money as they can out of their current investments (that largely involve MLC NAND). There's nothing to do about it; we're just going to have to be patient and wait for "business to run its course."
In the mean time, I've now been waiting 15 years for some form of
MRAM to finally come of age: a type of universal memory that will blow away all the existing forms of memory, both volatile and non-volatile. Unfortunately, I suspect that for many more years this type of memory will exist only in my dreams.
... it's reliability that's the real issue. SSDs are a great idea in theory, but in practice the only time I tried to build a server around one, taking great care to ensure that as little as possible would ever be ever written to it (e.g. turned off atime, while/var,/temp,/home etc. were located on hard disks), it ended up lasting only about a month.
I would love to replace my hard disks, arguably the most critical and vulnerable components of my computers, with SSDs, but only if they are more reliable in the first place, and can thereafter be regarded generally as an improvement.
One of the things I hated most when I was writing for Wikipedia was the anti-science attitude of many editors there. I wrote mostly articles on biological organisms and was a strong proponent of using scientific names for article titles. Common names are simply not unique, a fact that has resulted in many heated and pointless debates (i.e. Tiger vs. Puma). I figure WP should try to move beyond that and embrace the advantages of scientific nomenclature that biologists have known about for 250 years.
Most of the folks who were actually busy writing the articles agreed, but every time an attempt was made to change the policies, our efforts would be met with great resistance from people who simply did not know what they were talking about, let alone make any contributions of the kind. You could see from their edit histories that these people were bureaucrats: they produced very little content and an amazing amount of hot air. Yet, they have enormous influence at WP due simply to their dogged persistence.
In my view, the fact that more productive editors are now leaving as opposed to arriving is only partly explained by the low-hanging-fruit phenomenon. I, along with many others, was willing to take WP -- or at least my small corner of it -- to the next level, but the problem is that those bureaucrats simply don't share the same vision. When it comes to certain subjects that enter into their own realm of consciousness, it seems like they'd rather keep things looking like an expanded version of the old encyclopedia that their parents once bought when they were kids. It's completely at odds with Jimbo's original vision, but try telling them that.
As a result, the easy work has already been done, but anyone with the knowledge to do the hard stuff is quickly discouraged. I suspect most professional biologists don't even bother; a few of the ones I spoke to outside of WP had a low opinion of the site precisely because scientific names were not being used for article titles.
Finally, there's the problem of vandalism. Since I've left, no one has stepped in to keep an eye on the articles I wrote, let alone expand them in any meaningful way. The vandalism, however, is constant. Most of the obvious stuff gets reverted, but it's the subtle vandalism that is the most problematic. Unless you're a specialist, you just can't tell the difference. Either WP should start paying specialists to keep watch, or they should start try treating their own volunteer specialists with more respect. I've heard for years that WP v2 was supposed to solve a lot of vandalism problems, but so far it hasn't appeared.
... that's what Apple are showing themselves to be. If, occasionally, I feel interested in their innovative products, this kind of news is what always keeps me from acting upon it.
It seems that Apple has become too successful for its own good -- somethings that seems to affect (virtually?) all major corporations these days. At first they start out with a cool product and they're good at keeping their customers happy about it. Then they become a success and make a lot of money. But, almost inevitably the company gets sold and/or one or more assholes take over the helm. Assholes? Yeah, they're the type that understand that "It's the stockholders, stupid!" Therefore, if a decision to do something differently will probably make the company more money -- even if it's likely to piss off many customers -- then it's the right thing to do. Hell, they usually don't even care about breaking the law; normally they're careful that the potential fines don't exceed the resulting profits, but that's not always the case. Not that I'm accusing Apple of ever having done anything illegal, but I see no reason to trust them either.
Show me a successful company that has always put customer satisfaction before profits, and I'll show you a company that you can trust (at least until the next change of management). Actually, I'll bet that there are lots of them... just precious few (if any) that are publicly owned.
At $700, this looks like an overly-expensive and inflexible gimmick to me. For instance, what happens if I want to use 2.5" disks instead of the usual 3.5" inch disks that this chassis seems to support? With some minimal modding, I could fit in more than six, but not in this case. And what's that silly handle for at the top of the chassis? I can already feel the flared bottom edge of the case bumping against my shins.
Twenty years ago I knew people who had been pushed into the role of sysadmin because their boss thought they'd be good at it... based on the fact that they seemed to be better at operating the xerox machine than anyone else. The only glory in that was that the file server was the newest office play-toy and they got the "privilege" of operating it.
In 1997 I had a new colleague who had managed to switch from plain-old administration to network administration. He was so happy about it: "I don't have to worry about my job now, because I'm in IT." Of course, he was deluding himself. He didn't have any real responsibility anyway, but that was a time when consulting agencies were hiring (and were able to rent out) just about anyone willing to say they felt comfortable behind a keyboard.
Now, after having dealt with a few more managers and clients, I don't see how there was ever any glory in it. Today there's nothing new about file servers, or computer networks, or Internet connections -- every office has one. It's only miserable when critical parts break down.
Unfortunately, despite the fact that it's nothing new, IT remains a frustratingly abstract and unfathomable subject to almost everyone else, so they rarely follow through with your advice and it's never their fault when things break down. Years later they may finally see the light, take your advice and then marvel at the results, but it takes luck and patience to ever get that far. The legions of incompetent M$ sysadmins don't help the overall image of this profession either.
The bottom line is that I started to feel misunderstood and unappreciated years ago. As an independent consultant I don't earn very much these days because I'm picky about who I do business with, but at least this way I don't have to put up with the kind of people who would otherwise make my life more miserable than anything else. Peace of mind, or at least freedom from avoidable stress: that's definitely worth something as well.
Slashdot Mirror
That means I'll soon have 825M bps down and 1M bps up to look forward to.
M$ is all about their Windows/Office monopoly. On the PC. Period. For almost two decades now, the company's whole business strategy has really just been about maintaining that monopoly. It's what they've always done best.
Sure, they dabble in many other areas, but unless it soon starts to look like it has the potential to make them billions, just like Windows and Office, they quickly lose interest.
Innovation? Nope. They just buy other companies that look interesting, assimilate the knowledge, achieve relatively little as a result and then move on. Seriously, I can't think of anything unique that they themselves have ever come up with besides M$ Bob, which was a total flop, and perhaps a miserable animated paperclip. All of their other ideas have either been bought or stolen.
Seen in this light it's easy to predict when M$ will die: when the PC dies. Once their traditional monopoly has faded, M$ will be just another software company struggling with the competition... except that they never learned how to innovate -- they've never had to!! Oh, of course there are more people at M$ who know that and fear the inevitable, but they can't change anything. Their ship's course was fixed long ago and it can never be altered. Ray should have known that and avoided them like the plague, but I suppose he was just being stupid or greedy... or both. Within that company, however, he's hardly a unique case; just more well known.
My servers have become ever more complicated over the years: first Bind9, ISC-DHCP, NFS, Squid, Exim4, Spamassassin, ClamAV, CUPS, Apache2, PostgreSQL, Samba, Wide-DHCPv6, NTP/GPSD, and Asterisk to name a few, but now I've thrown out NFS and added Kerberos, OpenLDAP and OpenAFS (the fabulous Andrew File System). However, even though my Debian GNU/Linux server can easily handle this entire load, there are drawbacks to running all of these processes on a single physical server machine. The ones I see involve security and compatibility.
From a security standpoint, the problem is that if the wrong people ever manage to gain root access through some weakness in, say, Apache2 or Exim4, then they've got access to everything. As for compatibility, whenever the decision is made to upgrade the OS of a system that has only one OS, many if not all of the applications it runs also have to be upgraded. If any major problems are encountered, the ability to run certain applications may be lost, or it may even be necessary to turn back and reinstall the old OS, which may represent a terrible waste of time, money and effort. This might also be referred to as a continuity problem.
To address these issues, I've started working with KVM and Linux-VServer: the former is like a more traditional virtual machine, always requiring set chunks of memory and disk space for its VMs, while the second is like working with chroot jails on steroids. Despite all of their magic, however, both were easier to configure and maintain that I expected. So, as long as there's enough memory, it's now possible to set up server machines that start out running little more than KVM and libvirt-bin, along with the usual LVM tools for disk management. These physical machines can then be used to launch a number of KVM virtual machines, each of which can in turn run one or more Linux-VServer machines.
In the above scenario, it's possible to isolate many different applications -- especially those that are relatively vulnerable or need to be kept extra secure (e.g. Kerberos). If the bad guys do gain remote access to any of the more vulnerable applications, then it's more likely that the damage they do will be limited. In addition, upgrading becomes much less of a risk, since the virtual systems can in principle be upgraded individually; if one application does not upgrade properly, it doesn't have to hold the rest back. Even better, if you always keep a gigabyte or so of memory in reserve along with a fair amount of disk space, then you can always opt to replace older virtual systems as opposed to upgrading them; a good way not to burn your bridges behind you in case things go wrong. Finally, working this way means complicated upgrades will no longer have to be done all at once: they can be spread out over days, weeks or months. The only drawback is that you're left with multiple operating systems to maintain instead of just one per physical machine. But, that now seem like a small price to pay.
As for virtualization on the desktop, I've been using VirtualBox for the past so many years, but I have no idea how long that's going to last, because I don't think we can trust Oracle. I guess I'll end up using KVM and Linux-VServer for that as well. I suspect that it will be better to do that now while we have the choice, as opposed to later when we have none.
... since it happens:
* Once every day at 10 hours, 10 minutes and 10 seconds. But okay, that moment only lasts a second.
* On the 10th day of every month at 10 hours and 10 minutes, lasting sixty seconds.
* On the 10th hour of the 10th day of October every year, lasting one hour.
Today's date is only special because the 10/10/10-moment lasts an entire 24 hours, which is way more time than most people will care to dwell upon in its specialness. True, it only occurs once a century, but just remember that it can't hold a candle to October of the year 1010, which lasted an entire month and won't be seen again for another 9 millenia.
There are, of course, plenty more 10/10/10-moments for nerds to look forward to in the far, far future that will last a lot longer than that (a year, a century, a millennium...), but then we would have to assume that the creatures of this planet will still be hanging on to the Gregorian calendar after all that time, which is perhaps doubtful.
Perhaps weeds would be able to help us out here, since they grow quickly and don't seem to require any fertilizer. Needing only water, farmers could be subsidized to plant endless crops of some particularly fast-growing (genetically tailored?) varieties that would subsequently be harvested and buried. Perhaps burial would also help to prevent these oxygen-deprived organic masses from turning into sources of methane, which as a greenhouse gas is 25 times more potent than CO2. Or, maybe the methane could be trapped and burned to produce energy. This would produce CO2, but not be nearly as bad as letting the methane escape into the atmosphere.
... As a matter of privacy-rights, I think this judge is off his/her rocker. ...
... or corrupt.
Now imagine somebody else asking you these questions... you see how complicated this can become? Luckily, it doesn't have to be that way. All you have to do is license Slashdot's patented OS Startup Procedure and you'll no longer have to lie awake, agonizing about how to deal with all this complexity!
That's what I think my problem is. Recently, a very good friend of mine told me that if I installed OpenAFS on the systems of anyone who is dependent on me for support, he would not be able to help me with that if things go wrong when I'm not around (no, he has no interest in learning about it). Not that I was counting on his support, but that's still frustrating to hear. I worked very hard on getting OpenAFS to work, but I have to admit that, even within the Linux community, it's a lot more complex and far less popular than NFS.
Originally, I was concerned mainly with scalability and security. NFS gets very low marks in these areas, while OpenAFS scores very highly. However, I think my good friend's view has something in common with the position that many large organizations take in these matters: they just want to keep things as average and simple as possible so that they will always be less likely to suffer problems when it becomes necessary to hire or fire IT personnel.
This is frustrating for me. My IT career started when I was 30, it peaked financially seven years later, but nine years after that my technical knowledge is still growing. Hey, I love what I know now! Technically speaking, I feel like I can run circles around my former self. The problem is, most employers aren't interested in what I know now; only whether I'm up to speed on the latest software from M$. There's a difference here: where I grew impatient and decided to solve the basic problems myself, most employers apparently still prefer to wait for M$ to solve them instead.
It's ironic. One of the things that attracted me to IT in the first place was that it allowed for creativity. That's still true, but if you want a career as a system administrator, it looks like you don't want to get too creative.
At the time may career as a NetWare sysadmin was just taking off, so it was another six years before I made the switch to Linux, but for me Win95 marked the beginning of the end of my belief in proprietary software.
What you're worried about is having to deal with IPv6 addresses that are the product of stateless autoconfiguration. Those look daunting for sure, but luckily you don't have to use them if you don't want to (let alone remember them). Specifically, you can can also choose to configure your hosts with fixed IPv6 addresses, either manually, or (much better) with DHCPv6 stateful autoconfiguration. Then you can have IPv6 addresses that are about as easy to remember as with IPv4, except that you'll have more than you'll ever need and they can all be public addresses.
Once they figure out that there's money to be made from jiggering the Internet traffic flow through their networks, there won't be anything to keep all of the ISPs from doing the same thing.
The problem is that as concepts go, Net Neutrality is pretty abstract: if some of us even have colleagues who don't understand what's at stake, we can be sure that the overwhelming majority of ISP customers don't know or care. So, if it's not enforced by law, to expect any individual ISP to voluntarily treat all of its Internet traffic the same would be similar to expecting there to be some large theaters that would be be willing to play movies without showing advertisements. Of course, all those theaters will tell you that if they did that, their prices would have to go up, and that's true. However, the fact is that they never give us that choice, because A) they know that most movie goers don't care anyway and B) they know that the advertisers would not like the viewers to be given that choice -- better to keep things simple!
Yes, really small theaters often don't bother with ads, but that's because they don't sell enough tickets. Advertisers are only willing to pay theaters significant amounts of money if they can be convinced that the ticket sales are high enough. Below a certain threshold there's not enough money in showing ads, so theater owners will often try to increase their ticket sales by advertising that they don't show any advertisements.
In the same way, only small ISPs would advertise Net Neutrality because A) they aren't big enough to convince any significant content providers to make deals with them and B) they can't afford the necessary equipment anyway. On the other hand, in this case there's nothing to prevent a small ISP's upstream service provider from jiggering the traffic. And for that matter, if Net Neutrality were not required by law, where would the ever jiggering stop for sure?
In mid June I set up my latest server based on Squeeze with the expectation that it would go stable this summer. For a while I thought perhaps I had jumped the gun and would be stuck with a relatively unstable system for a longer period, but I guess not.
In particular, I'm happy with Squeeze because I could use it to get my Kerberos-OpenLDAP-OpenAFS system working on both the file server and workstations. Not that I've ever use any FOSS other than Debian for my server, but after my attempts failed to get the latest Ubuntu client to run the necessary client software for this (unfortunately) uncommon, but very capable distributed file system, I suspected the same Debian version for the workstation represented my best chance of success. And sure enough: it worked straight away! Ubuntu may have certain benefits, but it seems that if you want a desktop system that is a little out of the ordinary, Debian is still your best bet.
For a year I used Kubuntu, as well as the Linux Mint version based on it, for my workstation. I switched because KDE on Debian, of which I'm a big fan, had become so unstable in the spring of 2009 as KDE4 was being introduced. I'm grateful to Mr. Shuttleworth that I had this option. I was forced to move back to Debian because there are currently too many bugs in the Ubuntu packages that I need to support a distributed file system based on Kerberos, OpenLDAP and OpenAFS. This all works with Debian (lenny or squeeze), so I figure Ubuntu is just too focused on the desktop to care about it.
That's a pity for two reasons. First, I definitely had it easy for a while as far as the desktop is concerned. I've been back with Debian for a month now and there are still a number of rough edges to my desktop experience: I've spent far too much time adding missing functionality and trying to get it all to behave properly. It's such a waste of effort when you know that it doesn't have to be like that anymore. Linux Mint is so easy, even a relative noob can install it and have all kinds of basic desktop functionality running and configured in just a few hours.
The second reason is because Linux workstations deserve better file server support than just NFS and SMB/CIFS. Imagine an office building that will soon house 2.000 employees and being offered the opportunity to set it up with workstations and servers using only open source software. Would you feel comfortable doing that with NFS or Samba? I wouldn't. OpenAFS, on the other hand -- now that's a capable file system. I know that I would be able to rely on Debian and OpenAFS for the file servers, but I would also prefer a distro for the workstations that would likely result in the lowest number of help desk calls. I doubt that would be Debian, but it would be great if it could be something based on Debian. With OpenAFS and distros like Ubuntu, I figure we're almost there.
From this perspective, I find it really strange that so many long-time Debian users can be so hostile towards Ubuntu. It's not like anyone is forcing them to use it. IMHO, if it's so easy to use that it not only gives normal users the necessary confidence to make the switch from Windows, but also to fix (most of) their own problems afterwards, how can that be a bad thing? Furthermore, if the Ubuntu project continues to succeed where the Debian project has not, perhaps the latter should look to the former for a little inspiration every once in a while.
Or Xen, or KVM...
When my Sun VirtualBox application asked me to upgrade recently to version 3.2.6, I thought nothing of it; sure, always good to stay up to date, right? It proceeded without a hitch, although the only difference seemed to be that the branding had been changed from "Sun Microsystems Inc." to "Oracle" along with some corporate artwork. Was that all? Apparently not. Afterwards, several Linux distros refused to install as virtual machines: somewhere half-way, the process would report a generic failure. Strange, because I had installed the same distros many times before (at least the older VMs I had of them were still working). So, suspecting a bug, I downgraded back to v3.1.8 r61349 and the problem vanished. Pretty weird for what's ostensibly just a cosmetic change. At any rate, I feel this does not bode well for the future of VirtualBox and I've already been looking into the alternatives; perhaps something a little less dependent on some corporate overlord will be a safer choice.
Too many members of the general public are ignorant of science, what its basic tenants are, how it works, why it has been so successful and therefore why it deserves everyone's respect and attention, especially when scientists warn us about things like tornados, the AIDS virus, volcanic eruptions, tidal waves, lead based paint, and releasing too much greenhouse gas into the atmosphere.
Unfortunately, teaching people the facts about the universe we live in is difficult and expensive. But when society fails to educate its members sufficiently about science, to teach them to think critically, then the purveyors of disinformation -- typically organized religions and corporate marketing departments -- are always there to enlighten them with their own versions of the truth.
What can we do about this? First, never cut back on education. An enlightened society is an educated one and maintaining it as such is a endless task. Second, make education accessible to everyone at no cost. Three, we have to be hard on ourselves to ensure that our teachers and educational institutions continue to live up to the highest standards. Four... spend money on marketing facts that are both generally accepted by the scientific community and important to society.
How do we pay for all that? Higher basic taxes, I guess (it will eventually pay for itself), but perhaps also by levying a tax on top of what those purveyors of disinformation spend on advertising.
So much environmental damage has already been done and we're all still strongly addicted to fossil fuels. We're like a bunch of heroin addicts from whom going cold-turkey in basically unthinkable.
When comparing energy sources, what we already have is always cheaper, so why switch? And even if democracy were suspended in the USA, if other countries did not follow suit, Americans would soon have to admit that suddenly going green will have put them at a serious economic disadvantage.
There are no easy fixes. Millions may now be aware that we are collectively flirting with environmental disaster, but that makes no difference if billions don't care. What's needed is for the vast majority of the Earth's inhabitants to agree to give the climatologists the benefit of the doubt and to live on less for quite a while -- voluntarily -- in order to get things fixed.
Unfortunately, we live in a world in which most people strive to accumulate more wealth, but it's concentrated with 1% of the population (who own 95% of everything and are loath to part with it) while most of the rest live in desperate poverty and are only concerned with their own short-term survival (which often includes destroying their environment). Sometimes we cooperate, but mostly we just compete with each another.
I keep thinking that on the largest scale, we're no better than bacteria. I imagine a hypothetical colony in a petri dish, together with some algae: if they controlled their growth so as to strike a balance with the algae, they could potentially live indefinitely in their little closed ecosystem (assuming there was always enough sunlight). However, that would mean that the bacteria would regularly have to force themselves to go hungry and die... seemingly for nothing, because actually there would always be food around. It's much easier for a population to allow itself to grow unchecked, so in reality the bacteria would simply consume all of the algae as fast as possible after which their entire population would die off. Of course, the bacteria could form spores with the last of their strength, but that's not really an option for us.
No, I seriously doubt that humanity is going to fix this problem in an organized, rational and peaceful manner. The alternative? Nature will correct the imbalance for us. There will be war and famine for the vast majority, billions will die, but about 1% of the population will continue to live in relative happiness. Nothing new, really.
For a moment, I thought the same, but then I remembered reading somewhere that insects can only see ultraviolet the way they do because the light-sensitive cells of their compound eyes are so close to the surface. In the case of the mammalian eye, however, evolution has caused the lens to become opaque to most frequencies of ultraviolet light to avoid damage to the retina. So, even if the retina was sensitive more of the UV range, it would still not be able to detect it.
As for the possibility of seeing infrared light, if it were ever to become an option I'm not so sure that I would want to see too much of that range either. For example, it might be interesting to be able to detect objects and other living things by differences in temperature the way a pit viper does, but that would also have serious drawbacks. For example, a hot frying pan might become too painful too look at, and you'd also have to get used to falling asleep with the lights on, so to speak: after closing your eyes, you'd still see the light (the heat) from the inside of your own eyelids!
It's more likely though, IMO, that we will never be able to see infrared directly with our eyes. Snakes, for example, have evolved several different methods of detecting warm-blooded prey (parallel evolution), but those solutions all involve heat-sensitive patches of skin or scales (on, or between the labial scales of many boas and pythons, and of course the famous pit organ) -- it never involved the eyes. Other animals that see better than we do at night make use of a layer behind the retina, called the tapetum lucidum, that reflects light that has managed to pass undetected through the retina back out again, thus increasing the chances of it being detected on the second pass. For those who would be willing to pay for it, this might be a more interesting feature to add to the human eye, but it still would not help to detect infrared light.
Disclaimer: I'm an American, but live in Europe.
Right now the situation with health insurance in the USA is so awful that any change is welcome, just as long as that change does not actually come from the health insurance industry itself.
The problem with the health insurance industry in the USA is that it is hardly regulated at all. It's unforgivable that the law currently allows them to choose not to insure people based on their personal medical histories, and even worse that people who are insured (and pay the highest premiums of any developed nation) can be (and often are) given such a hard time when health insurance companies decide that they simply don't want to pay.
Still, so many people in the States seem reluctant to accept any new government proposals for change, because they fear that their already bad situation may actually get worse. They shouldn't be. The only ones who really have anything to fear are the health insurance companies, who are, incidentally, fighting the proposed legislation tooth and nail. That alone should allay people's fears, but amazingly, few seem to notice.
This coming weekend, if Obama fails to pass his health care plan, it will be a very sad day indeed for the American people. But, I won't blame Obama: as president, he can only do so much when congress refuses to cooperate. I want to see that Americans are still capable of helping themselves, but I fear they are not.
Let's face it, America has been a plutocracy for decades: it's run by the rich top percentile who cynically manipulate public opinion to their advantage... not hard to do in a country where most people have a pretty miserable education. Wake up, Americans: your country's star is fading fast.
Sorry, but I neglected to mention that the SSD in question was indeed both an early and a cheap model. I believe the manufacturer was OCZ, but I forget which model it was. This may subtract some weight from my original assertion, but I don't think all of it.
By the end of 2008 the Intel X25-M was supposed to be the best thing around, but even that model suffered from a form of low-level fragmentation that was the result of using both wear leveling and write combining. These are workarounds for problems that are inherent to SSDs that use NAND flash chips -- especially of the MLC variety -- and a lot of the competition involves trying to develop controller chips that are better at keeping that inevitable fragmentation under control. Only, I don't think any of them will ever find a practical way to really solve the issue once and for all. So, I'm getting tired of hearing about the "latest storage breakthroughs" that are based on NAND flash technology.
Okay, then why don't the chip manufacturers just ditch NAND all together and switch to a different technology with better prospects? Well, first because there's no demand for anything else (um, they don't want to invest the money that will create the demand), and second for the same reason the hard disk manufacturers don't want to quit: they first want to get as much money as they can out of their current investments (that largely involve MLC NAND). There's nothing to do about it; we're just going to have to be patient and wait for "business to run its course."
In the mean time, I've now been waiting 15 years for some form of MRAM to finally come of age: a type of universal memory that will blow away all the existing forms of memory, both volatile and non-volatile. Unfortunately, I suspect that for many more years this type of memory will exist only in my dreams.
... it's reliability that's the real issue. SSDs are a great idea in theory, but in practice the only time I tried to build a server around one, taking great care to ensure that as little as possible would ever be ever written to it (e.g. turned off atime, while /var, /temp, /home etc. were located on hard disks), it ended up lasting only about a month.
I would love to replace my hard disks, arguably the most critical and vulnerable components of my computers, with SSDs, but only if they are more reliable in the first place, and can thereafter be regarded generally as an improvement.
One of the things I hated most when I was writing for Wikipedia was the anti-science attitude of many editors there. I wrote mostly articles on biological organisms and was a strong proponent of using scientific names for article titles. Common names are simply not unique, a fact that has resulted in many heated and pointless debates (i.e. Tiger vs. Puma). I figure WP should try to move beyond that and embrace the advantages of scientific nomenclature that biologists have known about for 250 years.
Most of the folks who were actually busy writing the articles agreed, but every time an attempt was made to change the policies, our efforts would be met with great resistance from people who simply did not know what they were talking about, let alone make any contributions of the kind. You could see from their edit histories that these people were bureaucrats: they produced very little content and an amazing amount of hot air. Yet, they have enormous influence at WP due simply to their dogged persistence.
In my view, the fact that more productive editors are now leaving as opposed to arriving is only partly explained by the low-hanging-fruit phenomenon. I, along with many others, was willing to take WP -- or at least my small corner of it -- to the next level, but the problem is that those bureaucrats simply don't share the same vision. When it comes to certain subjects that enter into their own realm of consciousness, it seems like they'd rather keep things looking like an expanded version of the old encyclopedia that their parents once bought when they were kids. It's completely at odds with Jimbo's original vision, but try telling them that.
As a result, the easy work has already been done, but anyone with the knowledge to do the hard stuff is quickly discouraged. I suspect most professional biologists don't even bother; a few of the ones I spoke to outside of WP had a low opinion of the site precisely because scientific names were not being used for article titles.
Finally, there's the problem of vandalism. Since I've left, no one has stepped in to keep an eye on the articles I wrote, let alone expand them in any meaningful way. The vandalism, however, is constant. Most of the obvious stuff gets reverted, but it's the subtle vandalism that is the most problematic. Unless you're a specialist, you just can't tell the difference. Either WP should start paying specialists to keep watch, or they should start try treating their own volunteer specialists with more respect. I've heard for years that WP v2 was supposed to solve a lot of vandalism problems, but so far it hasn't appeared.
... that's what Apple are showing themselves to be. If, occasionally, I feel interested in their innovative products, this kind of news is what always keeps me from acting upon it.
It seems that Apple has become too successful for its own good -- somethings that seems to affect (virtually?) all major corporations these days. At first they start out with a cool product and they're good at keeping their customers happy about it. Then they become a success and make a lot of money. But, almost inevitably the company gets sold and/or one or more assholes take over the helm. Assholes? Yeah, they're the type that understand that "It's the stockholders, stupid!" Therefore, if a decision to do something differently will probably make the company more money -- even if it's likely to piss off many customers -- then it's the right thing to do. Hell, they usually don't even care about breaking the law; normally they're careful that the potential fines don't exceed the resulting profits, but that's not always the case. Not that I'm accusing Apple of ever having done anything illegal, but I see no reason to trust them either.
Show me a successful company that has always put customer satisfaction before profits, and I'll show you a company that you can trust (at least until the next change of management). Actually, I'll bet that there are lots of them... just precious few (if any) that are publicly owned.
At $700, this looks like an overly-expensive and inflexible gimmick to me. For instance, what happens if I want to use 2.5" disks instead of the usual 3.5" inch disks that this chassis seems to support? With some minimal modding, I could fit in more than six, but not in this case. And what's that silly handle for at the top of the chassis? I can already feel the flared bottom edge of the case bumping against my shins.
Twenty years ago I knew people who had been pushed into the role of sysadmin because their boss thought they'd be good at it... based on the fact that they seemed to be better at operating the xerox machine than anyone else. The only glory in that was that the file server was the newest office play-toy and they got the "privilege" of operating it.
In 1997 I had a new colleague who had managed to switch from plain-old administration to network administration. He was so happy about it: "I don't have to worry about my job now, because I'm in IT." Of course, he was deluding himself. He didn't have any real responsibility anyway, but that was a time when consulting agencies were hiring (and were able to rent out) just about anyone willing to say they felt comfortable behind a keyboard.
Now, after having dealt with a few more managers and clients, I don't see how there was ever any glory in it. Today there's nothing new about file servers, or computer networks, or Internet connections -- every office has one. It's only miserable when critical parts break down.
Unfortunately, despite the fact that it's nothing new, IT remains a frustratingly abstract and unfathomable subject to almost everyone else, so they rarely follow through with your advice and it's never their fault when things break down. Years later they may finally see the light, take your advice and then marvel at the results, but it takes luck and patience to ever get that far. The legions of incompetent M$ sysadmins don't help the overall image of this profession either.
The bottom line is that I started to feel misunderstood and unappreciated years ago. As an independent consultant I don't earn very much these days because I'm picky about who I do business with, but at least this way I don't have to put up with the kind of people who would otherwise make my life more miserable than anything else. Peace of mind, or at least freedom from avoidable stress: that's definitely worth something as well.