News from PANUG/BizNix - October 14, 2003
http://panug.org - http://biznix.org


SEALED SYSTEMS AT PANUG MEETING
The October PANUG meeting is Thursday night at 6:15 pm
at Novells office. Directions are here:
http://panug.org/meetings

The main presentation will cover sealed systems - a
revolutionary technology that has these benefits and
more:

o Easily install and upgrade computer operating systems
  and application software in minutes with few or no
  questions asked.

o Dramatically reduce the threat of attack by viruses,
  worms, human attackers, etc.

o Lock down user desktops so they cannot make changes,
  reducing user-induced technical support issues.

o Simplify system backups.

o Disaster recovery procedures are greatly simplified.

The presentation is suitable for non-technical people,
so you can feel safe in bringing your manager, which
you can do at no charge even if s/he isn't a member.
Please RSVP by replying to this message.


COMCAST DNS ISSUES
by Ed Sawicki - Accelerated Learning Center / Tailored Computers

Based on strange complaints voiced on the various DNS
mailing lists, it seems that Comcast customers may have
DNS problems under certain conditions. It appears that
Comcast is redirecting _all_ DNS queries to their own
DNS caches. This is OK for recursive queries but it's
a disaster for anyone running their own DNS servers,
caches, or proxies that make iterative queries.

If you're a Comcast customer and having DNS problems,
I'd appreciate hearing about it: ed@alcpress.com


CLOAKING DEVICE MADE FOR SPAMMERS
At last weeks PANUG meeting, Dick Pilz brought up the
subject of spammers spoofing addresses using a new
technique. Here's the article he was referring to:
http://panug.org/49


DIRE PREDICTIONS
http://www.fourmilab.ch/documents/digital-imprimatur/


ANTI-SPAM
by Jeremy Grand

There's alot of progress in the anti-spam war lately, at
least for client-based solutions. I tried the consumer 
reports-recommended product a couple of months ago, but
it just hung my PC. Spambayes seems to work really well.
I use Outlook, so I don't know how well it works with
other mail clients. It's an open source product available
on sourceforge.


FORMAT CHANGE
We're testing a new format for the articles that we
publish here. For articles where feedback is likely,
we're sending the article to members that we think have
expertise in the subject area in advance so their
feedback can be included with the article. Other members
are free to submit their feedback after the article is
published. The article by Ray Robert (below) is the first.


LINUX NOT A PERFORMER
by Ray Robert - System Administrator
Oregon Board of Medical Examiners

During a commercial database vendor's benchmarking of
their product on Linux vs. Windows they found that the
ReiserFS (running on Linux) on average handily
outperformed ext3. But this impressive average masked
wildly erratic performance, including several episodes
of 10 seconds of no response. JFS did slightly better than
ext3.

On the underlying benchmark on dual-processor systems, the
Windows (2000 Server) and Linux versions ran comparably.
They did not attempt kernel tuning, shutting off default
services, etc. It did surprise me that there was no
significant difference between Red Hat (7.3) and Mandrake
(9.0). I thought Mandrake's i586 kernel would be an advantage.

While there may be financial and security considerations,
the test suggests that there is (yet?) no performance
reason to migrate to Linux. The vendor's presentation is
at http://panug.org/47

Dick Pilz: Readers can go to http://panug.org/48 for an
article that discusses Linux file systems and why one file
system might be chosen over another. Criteria to keep in
mind are CPU load, space consumption, file size limitations,
disk throughput and file security methods.

Ken Barber: One database I/O benchmark test is an awfully
narrow basis for making generalizations about overall
performance.

Ed Sawicki: a)It's common for disk benchmark tests to
yield the same results with kernels compiled for different
processors. Most systems are blocked by the disk channel(s)
long before they're processor-bound.
b)Benchmarking systems is a terribly complicated process
to do right. Interpreting the results is harder still.
Discussions of the results are almost always religious.
c)Ray's conclusion may be correct. Linux might offer no
performance benefit over Windows. However, the security,
reliability, cost, flexibility, moral, choice, and freedom
benefits are hard to ignore.


DISCLAIMER
PANUG and BizNix welcome contributions from all members.
Member contributions do not necessarily represent the
official positions of PANUG or BizNix. The views of
members that contribute frequently may appear to be the
official position of the group(s). If you contribute,
you'll be adding vital diversity of opinion and outlook
to these broadcasts.