Slashdot Mirror

Incidentally, Snort isn't "SNORT" or "Snort!" or anything other than Snort. Snort isn't an acronym, it's an IDS. :)

Helevius

Thanks for playing the prior art game, play again!

Don't mind if I do. Implemented at the Univeristy of Alberta in the summer of 1999: Dealing with Public Ethernet Jacks - Switches, Gateways, and Authentication

Go to http://www.usenix.org/. Read their website. Join Usenix. Go to their conferences.

In particular, go to the LISA conference. It's all about Unix System Administration. You'll meet all kinds of people who know Unix. You'll attend all kinds of technical talks about Unix. And you can also sign up for various full/half-day tutorials before the conference.

If you want google to paw through all your files and risk having your account yanked for violating the user agreement, feel free to use it... (heck, maybe google won't yank your account in return for the opportunity to index your files...)

Mail-based file systems are nothing new, nor are http-based file systems (or WebDAV, for that matter).

The solution is to use an hash algorithm such as Expensive Key Schedule Blowfish. This allows variable cost in computing the hash such that the administrator can arbitrarily increase the difficulty of cracking his passwords without changing them.

Mac OS X relies on XNU (modified Mach 3.0 + BSD) for SMP. As far as I know, it uses a mutex model (like linux, freebsd, & netbsd).

And what about all of us geeks? Well unfortunately, I fear that Apple lose out here again.

Is that why about 50% of the laptops at the USENIX Advanced Technical Conference last week were Macs?

Seems like lots of geeks, at least the ones that go to USENIX (people like, er, Rob Pike, who might know something about innovative software) use Macs.

"In true Microsoft style, Redmond adopted a new metric for determining whether a CA's root certificate is to be included with its browser/OS/kitchen-sink product: In order for a CA's root certificate to be accepted - I swear I'm not making this up - Redmond said CA must pay a WebTrust-licensed member of the American Institute of Certified Public Accountants up to $250,000 for an initial evaluation/inspection, plus additional tens of thousands of dollars in fees on a periodic "follow-up" basis.

The makers of the Opera Web browser did not respond to email queries regarding their inclusion policies/requirements; however, a Bermuda-based CA representative stated in the netscape.public.mozilla.crypto newsgroup that "as of [his] last contact in 2003, Opera wanted cash to add a CA [root certificate]. They did not appear to have a standards policy.".

He goes on to describe the process of getting the root cert, hopefully, included into the Mozilla project through a Bugzilla feature enhancement request. From what I read from the article, the discussion about this is still going on.

Crappy slashdot anti-trollware.

Here's the right thing:

Peep - the network analyzer.

Here's the standards Eraser claims to satisfy. It uses US DoD 5220-22.M and Guttmann . The Guttmann method of erasing is only exceeded by using C4 as you decribed.

Technically even after a format (or several) the data can be still restored... Even if you overwrite it with random bits - reason being that the way hard drives work. Data is stored as bits encoded by small clumps of magnetically charged partials. These partials are arranged by polarity. But they are never really 1 or 0, think about them as more .7877 and .2314, if I write to the same 2 bits again with the same data the # will approach 1/0 respectably. There are hardware units that based on the actual charge can uncover layers of data. Full information is here.

Except that Softupdates isn't a journaling filesystem. Linux users have been brainwashed into believing that they need journaling.

For more info on softupdates versus journalling, see Soft Updates and Journaling versus Soft Updates.

Except that Softupdates isn't a journaling filesystem. Linux users have been brainwashed into believing that they need journaling.

For more info on softupdates versus journalling, see Soft Updates and Journaling versus Soft Updates.

From FreeBSD, that is... Would be nice to see that compared to Linux' FS-es. As in this earlier benchmark (PDF).

It's a well known fact that Leonardo was just a flamer who loved drawing homo-erotic pictures of naked men spread eagle.
And here is another one even more revealing of his tilt. What is less well known is that Leonardo invented sexual aids such as the french tickler. Here is his original sketch of the idea, which he called the Antarctic Tickler with a second sketch showing the device in its installed position. His bold innovation of using animal imagery in lovemaking culminated in the 20th century with gay men putting hamsters in each others asses.

There was a paper in USENIX 2003 about how things like email can be further reduced in size by delta-encoding similar files. It goes a step further than simply finding identical files or attachments.

It's cool to see different schedulers coexist; it looks like freeblock scheduling could supplement or improve anticipatory scheduling, for example.

I'd never buy a wooden computer.

I'd never even accept one if it was given to me.

Reminds me too much of Greeks Bearing Gifts.

(clue for the philestines: I think we have enough trouble with TROJANS already)

... All 20 of them.

... All 20 of them.

New Scientist is just carrying their little summary; one of the authors has the paper available on his site in HTML, PDF, and PostScript forms. It's to be presented at NSDI '04.

I think academic CS is majorly wrong in this regard. "System administration" is the residual work that is left over after you have abstracted all the easy stuff. Lather, rinse, repeat for 40 years, and system administration of very large & complex systems is highly distilled complexity and difficulty. Small wonder that senior system administrators are grand wizards of black magic within their field. I have published at LISA, twice. I get little credit for it, except among system administrators.

Crispin

I think academic CS is majorly wrong in this regard. "System administration" is the residual work that is left over after you have abstracted all the easy stuff. Lather, rinse, repeat for 40 years, and system administration of very large & complex systems is highly distilled complexity and difficulty. Small wonder that senior system administrators are grand wizards of black magic within their field. I have published at LISA, twice. I get little credit for it, except among system administrators.

Crispin

50GB are needed to compile the source, not just to host it.

took a week to setup and 2 hours a day to update

No wonder, with half a meg of memory :-)

Second point: The odds of getting one's hands on the full source to NT4/2K are slim to none--even most Microsoft folks couldn't do that. The code is probably scattered across multiple servers in Redmond, for starters, and you'd only be given access to the parts you needed to work with.

Microsoft gave a talk at usenix: Windows A Software Engineering Odyssey

This slide indicates the full source is 50gb and took a week to setup and 2 hours a day to update.

That implies to me that people could have the whole source but it would huge.

Slide 24 talks about their new perforce based system that only takes 3 hours to setup and 5 minutes to update.

Second point: The odds of getting one's hands on the full source to NT4/2K are slim to none--even most Microsoft folks couldn't do that. The code is probably scattered across multiple servers in Redmond, for starters, and you'd only be given access to the parts you needed to work with.

Microsoft gave a talk at usenix: Windows A Software Engineering Odyssey

This slide indicates the full source is 50gb and took a week to setup and 2 hours a day to update.

That implies to me that people could have the whole source but it would huge.

Slide 24 talks about their new perforce based system that only takes 3 hours to setup and 5 minutes to update.

Here is an interesting USENIX paper on FX!32.

Notable quote:
"After translation, x86 applications run as fast under DIGITAL FX!32 on a 500Mz Alpha system as on a 200Mz Pentium-Pro."

And that's after you've run the program a few times. The first time you run FX!32 is incredibly slow. Binary translation isn't easy stuff; there is essentially 0 chance that a 500Mhz alpha could match a 500Mhz PIII running any real x86 code, much less beat it.

"Timing the Application of Security Patches for Optimal Uptime". Steve Beattie, Seth Arnold, Crispin Cowan, Perry Wagle, Chris Wright, and Adam Shostack. Presented at the USENIX 16th Systems Administration Conference (LISA2002), Philadelphia, PA, December 2002. Postscript. or ugly PDF.

Crispin
----
Crispin Cowan, Ph.D.
CTO, Immunix Inc.

Actually, what everyone refers to as "Windows" now was actually introduced in 1993. VMS, which is still used (and will continue to be used) for many industrial control applications was introduced in 1978, and VMS itself was based on a few prior operating systems such as the RSX-11* family.

Unfortunately, it doesn't go online for about a month, and even then, you'll need a member password to view it online, but I imagine enough /. readers have access to ;login: to make it worthwhile to point this out.

Unfortunately, it doesn't go online for about a month, and even then, you'll need a member password to view it online, but I imagine enough /. readers have access to ;login: to make it worthwhile to point this out.

I don't know if the music industry would accept this as fair use, but it would certainly seem reasonable, much like a library lending out a copy but only for the duration of a given play. Perhaps it's a model for a new "legal P2P music sharing network": register CDs, ensure they can be played but not copied (perhaps with trusted client software), and prevent simultaneous play.

I'd sign up for that, and throw in my collection of maybe 1000 CDs.

One of the earlier works appeared in Slashdot, for instance here in 1999. But neither that column nor this hits for me on a search for military despite the military implications.

Specifically, there was a paper about this work in the 2000 USENIX Annual Conference. It mentioned detecting a loss of network connectivity during the NATO bombing of Yugoslavia during the period of their study, something the military could use to monitor the efficiency of their campaign.

dd if=/dev/zero of=zeroes

Or, if you don't want to mess with scripts and installing cygwin:

1. Download Eraser from here. (A very nice privacy tool for Windows, BTW.)
2. Install
3. Go to erasing prefs (Ctrl+E) and click New
4. Enter description: All Zeros
5. Click Add
6. Click Save
7. Select new "All Zeros", go to Unused Disk Space tab and do the same.
8. Click Ok
9. File - New Task (Ctrl+N)
10. Set up tasks for the drive(s) you want to zero out and then run them.
11. Profit! (Sorry.. couldn't resist.)

BTW.. if you want to use this for privacy, you probably *don't* want to use the All Zeros overwriting option. If you son't know why, read this interesting article.

• Jacal Paper Abstract (HTML)
• JACAL Paper (PDF).
• USENIX - LISA-NT 2000 presentation schedule

I'm terribly sorry for forgetting this. It's been three years since the conference and two years since I've touched JACAL at all (or worked at the university).

• Jacal Paper Abstract (HTML)
• JACAL Paper (PDF).
• USENIX - LISA-NT 2000 presentation schedule

I'm terribly sorry for forgetting this. It's been three years since the conference and two years since I've touched JACAL at all (or worked at the university).

• Jacal Paper Abstract (HTML)
• JACAL Paper (PDF).
• USENIX - LISA-NT 2000 presentation schedule

I'm terribly sorry for forgetting this. It's been three years since the conference and two years since I've touched JACAL at all (or worked at the university).

Better is Wayne's World 2, at the end where Garth meets the geek girl: "Is that a Unix book?" "Yeah." "Cool." (The book is the classic Steven's "Unix Network Programming".)

I've just written a short book on this. Documentation for SysAdmins, SAGE/USENIX's 11th short-topic book, should be available shortly (probably at LISA 2003.

Init Replacements

I say we just 'borrow back' SystemStarter from Apple/darwin.

It's got everything most people want and it's massively field tested. 'Not invented here' is its main problem.

I have started cutting these "Objectives" out and hanging them on my wall so I can look up occasionally and get a chuckle on an otherwise tense day. All job objectives tend to boil down to the following statement: "Seeking a position where I can work insanely long hours and use my godlike skills to solve impossibly complicated problems while working hand in hand with people who love me dearly because I'm such an amazing person."

Oh please. These types of statements are neither honest nor objective. More often than not, they are just plain silly.

For the most part, I personally feel that objectives and goals are a waste of space unless you really know what you want, and it is fairly specific. If you genuinely know, for example, that you are "seeking a management position in the technology field, preferably in IT, secondarily in development" or something like it, then great! Those people who have the job that you are looking for may call you. Those people who are looking for a peanut vendor will not. Perfection.

Actually, I'm surprised this isn't the de facto standard, at least in the commercial linux desktop distributions, since startup time is a reasonable thing to optimize for desktop machines (a journaled filesystem certainly helps :-). Mac OS X has a parallelized service startup sequence as well, and a dependency grouping technique - see this paper for some information.

Quote from the paper: For this reason it is effectively impossible to sanitise storage locations by simple overwriting them, no matter how many overwrite passes are made or what data patterns are written. However by using the relatively simple methods presented in this paper the task of an attacker can be made significantly more difficult, if not prohibitively expensive."

The program I use in Windoze to erase/overwrite files does 30 passes according to the principles set forth in the paper. If I were storing really sensitive info on it, I'd do as the Department of Defense does and physically destroy the media.

If you ever have to handle confidential information, depend on the info in the paper I linked to, not your guesses. This isn't a matter of research that needs to be done, this is work that was done years ago by people really motivated to find out the right answers.

Now, I've got to get an update for the program, which is oddly enough, called Eraser. Wish there were a Linux port.

I second the suggestion of submitting the work to a journal. Specifically, you could submit it to the LISA (Large Installation Systems Administration) conference, and get it into their proceedings. Or you could try ;login, the magazine of USENIX and SAGE.
Also, there's Sysadmin Magazine. If it's Linux related, there are a whole slew of Linux mags, of course.

xnu, the Darwin and OS X kernel, is actually monolithic. It just provides mach-like abstractions. Or something. See here (section 2) for details.

Mac OS X uses micro kernel technology. This provides better memory protection between applications, and the ability to sperate the OS into different components and levels. This becomes key when updating the OS. Most updates, since it does not involve the micro kernel, a complete system restart isn't necessary. The micro kernel will continue to run while the rest of the OS is patched in restarted, reducing start up time for kernel updates.

xnu is not a traditional microkernel as its Mach heritage might imply. Over the years various people have tried methods of speeding up microkernels, including collocation (MkLinux), and optimized messaging mechanisms (L4)[microperf]. Since Mac OS X was not intended to work as a multi-server, and a crash of a BSD server was equivalent to a system crash from a user perspective the advantages of protecting Mach from BSD were negligible. Rather than simple collocation, message passing was short circuited by having BSD directly call Mach functions. While the abstractions are maintained within the kernel at source level, the kernel is in fact monolithic.

introduction
WSD, it seemes to be allmost forgotten here in the civilised worlds but here and there you still need to bring support to some legacy systems as the WSD-2K/XP or the even more archane WSD-03
luckally there ARE ways to allmost be as productive in your new login shell CMD.EXE (aka Command Prompt or DosBox) as you where in your shell, *cough*

documentation
When you expect your now fairly standard info, man and share/doc documentation you will feel, at first exploration of the DosBox 'help' function (C:\>HELP or C:\CMD.EXE /?), as though you forgot to install some packages, feer not! this only appears to be this way, the list shown here is the equivalent of your once so feared, loved and reveared {s,}bin directories.
most of these programs can be invoked with the ? argument to get there --usage statement

help background
help started as a builtin in the olden dages but has since been replaced with an utility much like the GNU 'info' program.
having a CLI widget interface which supports hyperlinking, ANSI and more,....
(damn shame there ain't much to hyperlink to;)

argument invokation
Once accustomed to using backslashes as directory seperators you might look back at previous example and see the frontslash (/) used there as a hyphen (-) to pass an argument to a program
eg $ ls -R == C:\>DIR /S
or $ ls -h == C:\>DIR /?

shell grammar
CMD.EXE syntax does get more features every time they do the 'Even Better, Improved' thing ;)
i even think they are working hard to get it to one of those POSIX-Level 1 compliant shells!!
It has been around for ages, so who knows, maybe the marketing reps will say go, and once again someone removes the /* from various lines of precious future customer satisfaction that "the customer just don't want right now (tm)", and have support for an already replaced by v2 arcane standard that, imho,is hopefully deprecated and (pray to Hasthur) soon replaced by something nice oo-ish or something....
loops, variable substitution, etc is not available when you run cmd.exe interactively.
but it can also be used to run stand alone shell scripts who are called *.BAT files.
In these files you can, for instance, use a FOR (C:\>HELP.EXE FOR) that looks allmost like but then completely different as a 'bash' 'for' statement
the CMD.EXE /? man page has more information on these SHELL BUILTIN FUNCTIONS

Compatibility
thank god microsoft has this funny view on backward compatibility layers, this, in their perspective, has nothing todo with protocols and such but with leaving _everything_ on the system, anyone really did "START>RUN>progman.exe" or some other arcane 16bit app from the golden 3.11 ages on there sparkling new X-MS-OEM-Jailed box to really, really use it?
"if it hasnt broke, leave it alone,and whatever you do, never EVER try to hit it harded"

Next Week
How MS Uses a 3 char string for all its EXEC, MIME and other needs. aka, why 640k would've been enough.

to be fair; you ofcourse also have .js .vbs .cmd .bat to run scripts from,
but then, this also seems to be true for .doc .xls .* .** .*****
Regards and GOODLUCK
(hehe, You DID already ported the servers to *nix with ACL's and stuff right?)

Thijs

You have already contributed and are continuing to contribute.

O! Canada. Thank you.

Excellent article you linked to! I found this link from the article particularly interesting, in the context of the comments from the German fellow who saw SCO's "evidence" without signing the NDA. Here's a former SCO (now Intel) employee that worked on improving the Linux kernel for enterprise application (when he worked for SCO) talking about *scheduling* routines for *SMP* systems using Linux kernel 2.4.18.