Re: OS (Reliability,Security,Performance,Cost,..) Comparison And Hardware Reliability
From: DoN. Nichols (dnichols_at_d-and-d.com)
Date: 10/05/04
- Next message: Kris Kennaway: "Re: Hosed up ports systems"
- Previous message: Steve O'Hara-Smith: "Re: /dev/bktr ioctl help needed"
- In reply to: Keith Matthews: "Re: OS (Reliability,Security,Performance,Cost,..) Comparison And Hardware Reliability"
- Next in thread: Ted Unangst: "Re: OS (Reliability,Security,Performance,Cost,..) Comparison And Hardware Reliability"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: 5 Oct 2004 16:46:27 -0400
In article <W4udndZQSot87v_cRVn-hw@eclipse.net.uk>,
Keith Matthews <invalid@frequentous.co.uk> wrote:
>Gauth wrote:
>
>> Hy,
>>
>> I'm searching comparison (advantages, disadvantages) between
>> BSD,LINUX,WINDOWS.
>> What's the MTTF (Mean Time To Fealure) of this 3 OS ?
>> Why chose one instead of others ?
>> Does someone have objectives articles or book's name ?
>> Does someone have data on hardware : MTBF of the components ?
>>
>> I want to calculate the probability of failure for hardware and
>> software to choose the better OS.
>>
>>
>Looks like a troll or someone who doesn't understand what they are asking.
I'll assume the latter.
>Since the hardware will probably be identical you can ignore that.
At least, it *can* be identical. I'm running OpenBSD on both
Intel hardware (a Shuttle SS51G FWIW) and on two Sun SPARCS (a SS-5 and
a SPARC Ultra 1/140, and running Windows on the latter two machines is
not an option.
>As to the rest you'll probably find that MTBF (Mean Time Between Failures)
>is probably more a function of the application with Linux and BSD. Since
>you didn't specify which application (or which version of Windows) you'll
>get no meaningful answer.
Agreed. Even between versions of Windows, you get differing
uptimes -- and many applications can crash Windows -- especially mixes
of incompatible applications.
But -- as a trivial example, let's look at the Intel-based
Shuttle mentioned above. (OpenBSD 3.5)
======================================================================
curlmakr:dnichols 0:23 # uptime
4:26PM up 59 days, 4:25, 2 users, load averages: 0.13, 0.10, 0.08
======================================================================
Essentially, it has been running full time since I built the
system. (Admittedly, it does not get heavy use most of the time, but it
is used sometimes for heavy image-processing work, as it is faster than
any other machine I have at present.)
Compare this with the Dells using an older version of Windows
(Win 98, IIRC) which the FAA used to replace a set of unix boxen which
were getting old. The entire air traffic communications in LA (I
believe) was shut down in the middle of the day because the Windows
tends to lock up after 41.someodd days, and the builders put in a
program to lock up the systems intentionally to force a reboot (which
was *supposed* to be done once every 30 days as part of maintenance).
Someone skipped that step, and the whole bank of them locked up at once.
It took several hours to get them all rebooted and talking to each other
again. (For information on this, see a recent issue of the comp.risks
digest.)
To show that the uptime above is not a fluke, here are the
uptimes of three more of my machines:
======================================================================
ceilidh up 267+23:30, 0 users, load 0.05, 0.09, 0.11
izalco up 56+20:02, 1 user, load 0.74, 0.32, 0.20
popocat up 277+01:51, 0 users, load 0.06, 0.00, 0.00
======================================================================
Those figures are days+hours:minutes
Of the two longest ones, one is SunOs 4.1.4 (a rather old version by
now) on A SS-10, and the other is Solaris 2.6 on another SS-10. The
shortest one was rebooted when an additional bank of disk drives was
added, and the longest one would be longer, except that represents the
most recent time that a power outage has been long enough to require
shutting down the UPS before its batteries gave up.
So -- hardware, and OS stability is not a significant problem.
(Though it might be on some of the Windows machines built for speed in
gameing, and running at the edge of the maximum clock speed, so the
cooling is severely stressed.)
Windows can be crashed by wayward applications. On unix,
normally the *application* may crash, but the OS remains alive,
especially if the application is not running with superuser privileges.
Some of my SS-5 machines do have spontaneous reboots, associated
with marginal memory speed, I think. (When I pay $5.00 each for 64 MB
SIMMs, I can expect some to be marginal. :-) These are typically after
two or three months of operation.
Just some anecdotal information.
Good Luck,
DoN.
-- Email: <dnichols@d-and-d.com> | Voice (all times): (703) 938-4564 (too) near Washington D.C. | http://www.d-and-d.com/dnichols/DoN.html --- Black Holes are where God is dividing by zero ---
- Next message: Kris Kennaway: "Re: Hosed up ports systems"
- Previous message: Steve O'Hara-Smith: "Re: /dev/bktr ioctl help needed"
- In reply to: Keith Matthews: "Re: OS (Reliability,Security,Performance,Cost,..) Comparison And Hardware Reliability"
- Next in thread: Ted Unangst: "Re: OS (Reliability,Security,Performance,Cost,..) Comparison And Hardware Reliability"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
