[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: why Linux?

> For instance, there was the Mindcraft benchmark test where NT blew linux
> 2.2 out of the water, and then Dave Miller's response in the 2.4 kernel
> bringing the upper hand back to linux, network-performance wise.  Where
> is all this documented?

I don't know the exact link, but I have read articles which highlighted
that the Mindcraft tests were conducted on the Microsoft compound, and
paid for by the Microsoft team. Specifically, the test targetted a certain
configuration which was a known issue with the 2.2 series kernels that
caused a bottleneck in the TCP/IP stack. This uncommon bug was fixed in
2.4, which is why the later tests were successful. Furthermore, the test
was based on a specific web server setup (high load, static pages), which
offers a conclusion more that Apache shouldn't be used in such a
situation, not Linux.
> There are plenty of Linux advocacy articles out there telling you how to
> tell everyone else that Linux is Really Really Good without flaming
> them, but where are the hard data?  Not even the Linux International
> site seems to have any links demonstrating *why* Linux is preferable to
> NT.

http://www.jimmo.com/Linux-NT_Debate/ gives some good
arguments. Personally, I feel the best arguments are those of the
mindset. A Linux server will be more stable under unforseen situations
(for example, everyday life), and any problems that do develop will be
dealt with immediately, with less trouble. You have the guarantee that
bugs found in a Linux application will be documented and worked on almost
immediately after being found. I cannot endorse an NT server with the same
You're absolutely right that there aren't many facts out there proving
Linux is better. However, even if there were conclusive proof that NT
performed better than Linux, I would still recommend Linux first. This is
the exact argument I have used with different people to convince them to
choose Linux:

Linux and NT both have bugs, and both have bugs that have not yet been
found. When troubles do arise, and they will, do you want software by a
company which denies that a problem exists and ignores it until they can
charge at least $100 to fix it a year later, or software which will be
acknowledged and usually fixed within a few weeks, distributed for free?

By the way, I administer both Linux and NT servers. I am currently in the
process of converting the latter.


Jonathan Eisenstein

PGP Public Key: http://www.mindspring.com/~jeisen/pgp.asc

Reply to: