Re: How workable is Vista?

Vladimir Vassilevsky wrote:



A computer is not more of a fire hazard at night than during the day - less so if it is not doing much work.

At night, the computers are unattended. Long while ago I had to extinguish a burning TV set with water. Since then, the techology was improved etc. etc. but nevertheless.


It's certainly possible to be unlucky. But the chances are that any overheating that could cause a fire risk will happen during use, not at night. Overheating has less chance of turning into a fire (because you turn the thing off when you first see the smoke, rather than after your smoke detector has seen the smoke). Still, the risks are minor.

The fan noise is not a problem unless it is in your bedroom,

And the fan speed is adaptive also, but fans still suck the dust :)

and the hardware wear is not a big issue

Mainly because of the high rate of the moral depreciacion. However the progress seems to slow down despite of the evil efforts of M$ and Intel :)

(hard disk wear is much more significant when they are in use), though it is not negligible.

I heard the opinion that the hard disk wear is the most significant at startup and shutdown. Don't now how much truth is in it.


I heard that too - I also don't know the truth of it.

One customer I knew had a hard disk that was fine when running, but had to be held at an angle of about 20 degrees to the horizontal to start up, and shaken slightly at specific parts of the boot up sequence. Fortunately it ran a dedicated program on DOS, and only needed booted if there was a power cut.

Good quality hardware is designed to work at 100% uptime.

I forgot what good quality means. You can buy a cheap part or an expensive part, and there is about the equal likelihood of failure. Only the things which were produced for many years and in the big quantities can be polished to perfection. This is the opposite to the current demands of the market.


There's a lot of truth in that - there's a reason you can buy new servers equipped with, for example, 73GB SCSI disks - they are a few years behind modern SATA disks in space, but are tried and tested. But there are certainly differences in the care and attention to detail that come with better quality parts.

The only serious issue is energy usage. Modern systems are quite good at running at lower power when there is not much demand for processing power. It's still a waste, unless you use electric heating anyway. How relevant that waste is depends on your circumstances.

Wasting is wrong in the phylosophical sense :)


That's true. But if you do the maths, you'll probably find that you save more money, and reduce your "environmental footprint" more, by cycling to work one day a week instead of driving.


The Microsoft 9x family was quite useless for any serious
applications,

Like what?

A typical bashing of M$ without any reason :)

Bashing MS is fun and easy - is any other reason required?

As if it could help any :)


Most of the problems attributed to Microsoft are actually caused by the lousy application programmers and unskilled operators.


Most of the problems people encounter are application problems, or driver problems. But MS is not blameless - both applications and drivers can cause far more problems than should be possible if the OS had proper separation of programs.

BTW, this is the valid argument for change to Vista.


Unless Vista has moved the gui and other large and unreliable parts out of the kernel, made all third party drivers non-kernel (or at least made it possible to run drivers in user mode), separated the gui from the file manager, and separated the web browser from the rest of the system, then I don't see a reasons to change.

I read somewhere that it's now possible (or will be "real soon now") to disable the gui altogether on server versions of windows - almost like on a real server operating system. I haven't used windows as a server since an old NT 4.0 machine (that is still going strong), so I haven't followed the details.

The main reason NT 3.51 was more reliable and stable than any windows version before or since is that the gui and drivers were in user space, not kernel space.

But there are the performance and the compatibility tradeoffs associated with that.


Certainly you can get more performance with the gui and graphics drivers in the kernel (compatibility should have been a minor concern, as there were no application level compatibility issues), and that's why they were moved to the kernel. But no one complained about the speed of NT 3.51 - they were astounded by its reliability.

There's no doubt that having X and the graphics system outside the kernel on Linux makes X slower than Windows for high speed graphics on the same hardware. There is also no doubt that it makes the system more reliable (if your Linux desktop dies, the average user will still think "my computer's crashed" - but it comes up again faster, and you don't have the risks of file system corruption and other serious failures from a complete crash). With more modern Linux systems, however, the graphics system gets a limited back-door to the hardware - enough to get a similar speed to Windows for graphic-intensive games, without the risks.


The problem with any Microsoft version is that it needs daily (or at
least weekly) security updates,

There is no practical need to do updates to the original versions. Well, with the exception for the few special cases when you really need never components. It is Microsoft who wants you to be addicted on the unnecessary clutter.


I agree entirely. The only people that really need regular updates and service packs are those with unsafe web habits.

The unsafe habbits can't be helped. The non-technical problems can't be solved by technical means.


The unsafe habits can be cured in the office - threads with wire cutters and public humiliation have a marvellous effect on sinners. But as you say, it's a non-technical solution!

Use a proper browser (anything but i.e.),

We can like it or not, but IE is the standard. If you don't want to have strange problems with some of the web pages, you have to keep IE.


We don't like it, and it's not the standard. Usage varies widely according to geographic area and the sort of web site used for surveys, but the market is split roughly evenly between IE6, IE7, and Firefox. If you don't want to have strange problems with websites, you use Firefox. Apart from testing websites that I have made myself, I've *never* (in over ten years) used IE seriously for more than a small handful of sites in total.

But if you *do* need to use IE for some sites (I know some bank web sites have the ridiculous idea of using ActiveX for "secure" logins), it's easy enough to lock IE to "high security" in general, and specify the required site as a "trusted" site. Use Firefox (or Opera) for everything else.

a decent email client (anything but o.e. or o.),
proper virus scanning on email (on the email server, not the PC), and a real firewall (not software "firewall" on the windows machine). And apply a touch of common sense.

The common sense is the main component. With common sense, the IE and OE are OK. Without common sense, nothing will help.


No, common sense is not enough (though it is necessary). The difference is when you do something accidentally - click on an advert, mistype a web address, or fall for a phishing link. With IE, when the adverts for casinos (or worse) start popping up, and the "helpful" search toolbar is installed before you can close the page, you are stuck with it until an expert can (perhaps) fix the machine. With Firefox, you just close the page, or perhaps the entire browser, and you're fine. Even if you manage to totally mess up Firefox with such junk (I've never heard of such a thing happening), you can just uninstall Firefox, delete the profiles, and re-install - it's just an application. With IE disasters, you have the joys of re-installing windows to look forward to.

These basic rules are far more relevant than windows upgrades or fixes.

Exactly.

which strangely also require a reboot.

This is not strange. It is quite difficult to unninstall the loaded component correctly, with all associations and dependencies. They ask for reboot to make sure everything is done all right. In the most of cases you can continue safely without reboot even if they required it.

It's difficult, but far from impossible. On *nix, it is perfectly possible to replace and update the executables and libraries for programs that are in use. The new version will not be used until it is restarted, but that alone avoids a lot of reboots for equivalent updates.

This is what Windows does, too.


No, windows will not allow you to update a file that is in use. When a program is running that uses a file (such as the program file itself, or its libraries), the file is locked. Any directories seen by running programs are also locked and can't be renamed or moved. That's why installation programs always advise you to close all other programs, and often complain that you have to close specific programs before proceeding (they are using shared libraries). The nearest Windows allows to replacing live files is to leave a note in the registry that specific files should be replaced (or deleted) during the next boot up.

For a more advanced solution, there is work under way to allow live patching of the Linux kernel without rebooting, as long as the change does not require changes to existing data structures. It's already perfectly possible to unload, update, and reload kernel modules without a reboot.

On windows, many of the "required reboots" are not required at all - it's just developers don't spend the time and effort required to figure out how to avoid it. A reboot is simple to request, and avoids many potential complications.

Yes. It is difficult to blame the developers for attempting to make their lifes easier. BTW, what is so wrong with rebooting? It only takes a minite or two, if the PC is not cluttered.


I agree about not blaming the application developers, although in many cases they could have avoided a reboot with minimal effort.

But rebooting *is* time consuming for many users. I regularly have over twenty programs or windows open at a time. I have several text editors from work on different programs, or different parts of the same system. I have several Firefox windows, each with multiple tabs, with different reference pages conveniently on-hand. I have a bunch of command prompt boxes, ssh clients, explorer windows, and other stuff at hand. I don't reboot often - it's a waste of time and an inconvenience.


For this reason, I do not recommend to my customers to use any Windows
based operating systems, unless a double/triple redundant system is
used (in which case you can boot one PC at a time for security
updates).


If someone is a lamer, he will have problems regardless of OS.


If you need close to 100% uptime, you need redundant systems. With windows, you need more to achieve high uptime figures than with *nix. I can't remember where I read it, but I saw a report for getting five nines uptime for a cluster running a web server. With Red Hat, you needed two machines (obviously with appropriately spec'ed hardware - redundant power supplies, fans, and RAID). With Windows (probably W2K or W2K3 server), you needed five machines.

I'd say it depends mainly on the qualification of the personnel and the way they take on the things. There are the numerous examples of the system crashes at the internet providers and web hosters; regardless of what software and hardware they were using.


Incompetence trumps good hardware and good software every time!

There are also the numbers to consider - at five nines reliability, you have an expected average of five minutes downtime per year. But if you have a thousand systems at five nines, you can expect that for about 4 days of the year, at least one will be down. The probabilities get messy, and failures on single systems can lead to failures on others. All in all, it is not surprising that big players have total failures on occasion. *You* may have little chance of winning the lottery, but it's not surprising that week after week, *somebody* wins.
.

Relevant Pages

  • Re: How workable is Vista?
    ... clicks almost the moment it boots up; I don't have to wait for all the startup applications to finish before I can get responses to anything I do. ... The main reason NT 3.51 was more reliable and stable than any windows version before or since is that the gui and drivers were in user space, ... There is no practical need to do updates to the original versions. ... In the most of cases you can continue safely without reboot even if they required it. ...
    (comp.arch.embedded)
  • Re: How workable is Vista?
    ... reliable and stable than any windows version before or since is that the gui and drivers were in user space, ... There is no practical need to do updates to the original versions. ... In the most of cases you can continue safely without reboot even if they required it. ... unless a double/triple redundant system is ...
    (comp.arch.embedded)
  • Re: Complaint about update system
    ... Allow the user to postpone a reboot after a Windows ... should also be noted that this automatic installation is done at a specified ... Change Automatic Updates back to Option ...
    (microsoft.public.windowsupdate)
  • Installed Security Updates: now boots & dies (Gray screen of death)
    ... I had to shut down last time (vs. reboot) and, ... Anyway, today I try to start it up and it boots, shows the Windows XP ... Pro Black splash screen with the little 2" bar in the middle bottom, ... of updates to load... ...
    (microsoft.public.windowsxp.general)
  • RE: Stuck in a crash/reboot cycle
    ... I'm always suspicious of video driver updates. ... I've run into a bit of a problem after installing the latest round of ... Firstly, after they installed, I was busy so I didn't reboot immediately. ... Windows I kept getting a Windows Installer window popping up saying ...
    (microsoft.public.windowsupdate)