photo-rbt.jpg (2942 bytes)

Email Robert

Daynotes Journal

Week of 4/19/99

Sunday, April 25, 1999 09:10

A (mostly) daily journal of the trials, tribulations, and random observations of Robert Bruce Thompson, a writer of computer books.


TTG Home

Robert Home

Daynotes Home

Search TTG

Special Reports

Last Week

Next Week

Monday

Tuesday

Wednesday

Thursday

Friday

Saturday

Sunday

Top


[Monday]   [Tuesday]  [Wednesday]  [Thursday]   [Friday]  [Saturday]   [Sunday]


Monday, April 19, 1999

If you didn't read the updates last weekend, check back to last week. I posted quite a lot of interesting new stuff Saturday and Sunday.

* * * * *

Back to the grind. The priority this week is to take a couple of the chapters Jerry has worked on, reformat them, add some additional material, and get them back to Jerry for his final first pass. Then they'll go to O'Reilly as a first draft submission. Barbara has been working for the last week or more abstracting all of Pournelle's View pages, extracting war stories and so on, and pasting them into the chapter outline. He'll take the best of them, clean them up, add details, and insert them as object lessons in the text. This morning, Barbara starts on my Daynotes pages, working forward from 6/15 last year through the current one. After that, she'll start on Pournelle's Mail pages. By the time she finishes all this, she'll probably be sick of both of us.

* * * * *

Elliotte Rusty Harold [elharo@metalab.unc.edu], who also writes for O'Reilly, posted an interesting article on his web site, Why I Like Writing for O'Reilly. It pretty much sums up my experiences with O'Reilly as well.

* * * * *

More from Bo Leuf on digital cameras:

That would be nice, but I wonder how well it would work. In my experience, things intended to modify a product to do something other than what it was originally designed to do are never completely satisfactory.

True enough, but a CCD plate in the film plane matching a 35mm frame... not much modification involved, just replaces a normal film, so the only issue is the quality of the device itself. I have to track down the reference and follow up on what is happening.

Another item on my wish list is a good slide/negative scanner. There are a couple of decent ones on the market now, so I expect when the money starts rolling in I might go with one. I have a lot of snaps and slides I would love to work with digitally.

/ Bo

--

"Bo Leuf" bo@leuf.com
Leuf fc3 Consultancy
http://www.leuf.com/

Well, it would obviously involve more than just a CCD sensor in the film plane, since there has to be provision for processing and storing the data you capture. I'm sure that a removable back could be produced for professional grade cameras, but as a low-volume product it would likely cost more than a purpose-built camera. And I agree that a slide/negative scanner would be a handy item, but it's not on my list of priorities at the moment.

 

 


TTG Home

Robert Home

Daynotes Home

Search TTG

Special Reports

Last Week

Next Week

Monday

Tuesday

Wednesday

Thursday

Friday

Saturday

Sunday

Top


Tuesday, April 20, 1999

Congratulations to my wife. Barbara is now officially a professional writer. She's going to do a short article for Library Journal, and they're going to pay her for it.

* * * * *

I was just sitting here in my office looking around when a thought struck me. I started to add up what's here on my network. The machines on my home network cumulatively have about:

  • 50 GB of hard disk space
  • 1 GB of RAM
  • 2.5 GHz of processor

Then I got to thinking about what was here 15 years or so ago:

  • 0.01 GB of hard disk space
  • 0.00064 GB of RAM
  • 0.00477 GHz of processor

So in the intervening 15 years, the computing power available to me has increased by a factor of about 5,000 for hard disk space, about 1,600 for RAM, and about 500 for processor. Reversing the process and projecting that out 15 years in the future, that'd put my home network in the year 2014 at about:

  • 250,000 GB of hard disk space, or 250 terabytes (TB)
  • 1,600 GB of RAM, or 1.6 TB
  • 1,250 GHz of processor, or 1.25 THz.

Right about then, I figure we'll probably all be running Pentium VIII systems and still be waiting for Microsoft to ship the latest version of NT. Or we may all be running Linux by then.

* * * * *

This from Bo Leuf:

This is snipped from a recent issue of Risks Digest (20:31):

From: "Robert David Graham" rob@netice.com
Subject: favicon.ico

In case you haven't heard, Microsoft has a new feature in IE 5.0 web browser. When you add a website to you "Favorites" (aka. Bookmarks for you Netscape users), the browser attempts to download a graphic called "favicon.ico", then show that icon along with the title of the webpage.

This has two risks.

First of all, the website owner is notified when you the page to your favorites, revealing information about yourself. A discussion of this can be found at

http://msdn.microsoft.com/workshop/essentials/versions/ICPIE5.asp

This privacy risk is probably minor, but I've seen several press articles on the subject.

The second RISK is much more severe. Go to AltaVista (or any search engine) and search for "favicon.ico". You now have a list of 500 websites that expose their access logs. In the logs, you can find several websites that expose the URLs of CGI scripts, including passwords. Through manual searching, I found 2 sites that exposed logon information; I'm sure I can write a program that would scan those logs to look for CGI programs and get even more. This also exposes even more privacy information because these logs often contain the Referer field as well.

(Risks archives at e.g. http://catless.ncl.ac.uk/Risks/)

/ Bo

--

"Bo Leuf" bo@leuf.com
Leuf fc3 Consultancy
http://www.leuf.com/

That's interesting, although I'm not sure how much of a real threat it is.

* * * * *

The following message is from pair Networks tech support in response to a message I sent them about problems I was having with email. Although it took a few days, I was impressed with the detail of their response. So far, although I hate pair Networks' billing department, the service itself and their tech support staff seem to be good:

I've been having sporadic problems with email ever since I started using your service. I POP directly from your mail server for my "thompson@ttgnet.com" account and for general messages delivered to any *.ttgnet.com addresses. I also have several autoforwards set up, including one that forwards messages for "barbara@ttnget.com" to "thompsrb@bellsouth.net".

I frequently receive messages in my mailbox that should have been autoforwarded to "thompsrb@bellsouth.net", particularly return-receipts intended for barbara. That's bad enough, but what's worse is that messages addressed to both "thompson@ttgnet.com" and "barbara@ttgnet.com" are usually (perhaps always) delivered only to barbara@ttgnet.com, and never make it to my mailbox.

Do you have any idea what's going on? I've never seen a mail system behave so oddly.

This is the correct, though not necessarily preferred, behavior of mail delivery under sendmail on our servers. Needless to say, we are planning to implement qmail in place of sendmail as soon as possible.

The first situation you mention is fairly standard and is done to maintain system integrity. Messages that are deemed to be system-generated are not forwarded because they can create mailing loops. Such messages include mail from usernames such as root, admin, postmaster, and mail. There's not really a decent, safe way to forward such messages under procmail with sendmail.

The second situation you mention is caused by sendmail and will be corrected under qmail. You are encountering one of the limitations of procmail under sendmail, which is that it acts on the first matching criteria it receives. In this case the only address it sees is the first one that it encounters in your .mailproto file and only the first recipient receives the message. There is not much that you can do using only procmail to solve this; workarounds will generally make the recipients receive multiple copies of each message and are not very good. This should be better under qmail, which should be implemented fairly soon.

You may wish to consider the use of an alternative mail handler such as mdforward (see http://nemeton.com.au/sw/mdforward/).

Regards,

--

pair Networks' Support Department           support@pair.com
The Support Forum, online resource center: http://support.pair.com/

 

 

 


TTG Home

Robert Home

Daynotes Home

Search TTG

Special Reports

Last Week

Next Week

Monday

Tuesday

Wednesday

Thursday

Friday

Saturday

Sunday

Top


Wednesday, April 21, 1999

Well, FrontPage 98 just did it to me again. I fired up FrontPage Explorer, and the local copy of the root web took at least a full minute to load. This time, I suspected immediately what was going on. Sure enough, when I changed to All Files view, nearly every HTML file in the entire web had been updated to the current date and time. Nothing had changed since I updated this page yesterday that might have caused this. FrontPage is simply a buggy application. The last time this happened, I speculated that it might have something to do with the change to Daylight Saving Time. Apparently not, since it didn't have that excuse this time. I hate FrontPage. If I didn't have so much time and effort invested it it, I'd change to something else immediately.

* * * * *

After what happened in Colorado yesterday, my award for tastelessness goes to The WB network for airing the previously-scheduled episode of Buffy The Vampire Slayer, whose plot included a bunch of misfit former students invading the high-school and planting a bomb. What were these people thinking? I hope that at least the local affiliate and cable systems in Denver had the decency to black out that episode.

To state the obvious, something is seriously wrong with our school systems and with society at large when such things can occur at all, let alone as frequently as they've happened over the last year or two. I don't have children. If I did, they would not be attending public schools. Outrages like this seem to me to be due to the Law of Unintended Consequences. Instead of trying to figure out what actions to take to "fix" the problem, those responsible should be asking themselves what they've been doing all these years that allowed the problem to arise in the first place.

This is not solely a problem of the schools and teenagers, as evidenced by frequent similar incidents in postal facilities, by drive-by shootings, and by the frequent cases of young children torturing and killing even younger children. This is a structural societal problem, and quick fixes aren't going to solve it. Old values have been discarded, and the feel-good Politically Correct dogma that has replaced them is at the root of these outrages. Children used to be taught that behaving antisocially resulted in quick, certain, and painful punishment. Spare the rod and spoil the child. Nowadays, when parents may literally be jailed for spanking a misbehaving child, is it any wonder that children grow up believing that they can behave however they wish?

For many years now, children have been taught that misbehaving or injuring others will result in at most a slap on the wrist. Our courts do nothing to punish children who torture and murder. A child in its natural state is a savage, capable of incredible cruelty without feeling remorse or even understanding the nature of his actions. Children are not born with any instinctive moral compass. They must be taught what is and is not acceptable behavior. If they are not taught acceptable behavior when very young, they will never learn.

Much attention focuses on salvaging such sociopathic children. But the sad reality is that such children are unsalvageable. Once bent they remain forever bent. It's too late for them. We should recognize that fact and focus our attention on making sure that we don't turn out any more of these creatures.

 


TTG Home

Robert Home

Daynotes Home

Search TTG

Special Reports

Last Week

Next Week

Monday

Tuesday

Wednesday

Thursday

Friday

Saturday

Sunday

Top


Thursday, April 22, 1999

I noticed this morning that both Governor Jesse Ventura and Jerry Pournelle were courageous enough to say something that I'd intended to say, but didn't. The "zero tolerance" weapons policies so popular nowadays almost certainly contributed to the high body count. Had even one of the teachers or administrators been armed, the situation could have turned out very differently, perhaps with only the goblins dead.

Some might argue that an armed teacher would have had no chance against these students. After all, there was an armed sheriff's deputy in the building, who shot and missed. If that deputy failed, how could a civilian have succeeded? Well, look at it from the goblins' viewpoint. They walk in the front door as wolves, and see a bunch of sheep. The sheriff's deputy by virtue of his uniform is obviously a wolf, someone to be disposed of before getting to work on the sheep. In fairness to that deputy, one guy with a handgun is no match for two with shotguns. But that armed teacher is a wolf in sheep's clothing. The best way to win a gun fight is to shoot the other guy before he even realizes there's going to be a fight. By virtue of his sheep's clothing, that teacher would have had a decisive advantage over the goblins.

Twenty years ago and more, I was in Rhodesia on a brief visit. At the time, Robert Mugabe and Joshua Nkomo led opposing terrorist factions that frequently engaged in outrages against civilians. The morning paper reported an interesting story. Several terrorists armed with AK-47s had hijacked a bus, intending to kidnap or murder the riders. They immediately murdered the driver and one or two soldiers in uniform. One of the riders, an elderly woman, pulled her pistol from her purse, stood up, and started shooting. She wounded at least one of the attackers, and drove the rest away. Had that woman not been armed, it's very likely that everyone on that bus would have died that day.

Predators are confused when the prey fights back. All that gun laws accomplish is to make sure that prey remains prey. Gun laws disarm only the good guys. This is insanity.

* * * * *

This from Bo Leuf [bo@leuf.com]:

Bob, you wrote...

"...A child in its natural state is a savage, capable of incredible cruelty without feeling remorse or even understanding the nature of his actions. Children are not born with any instinctive moral compass. They must be taught what is and is not acceptable behavior. If they are not taught acceptable behavior when very young, they will never learn...."

I disagree with this "natural savage" viewpoint, but am aware that this moral and ethic issue has been debated both ways for thousands of years. At least. Societies rules may sometimes be arbitrary, sometimes clearly for the greater good; the important point is that there are rules that the young must (learn to) follow until they become responsible adults.

I think you are more on the mark in the earlier comment "For many years now, children have been taught that misbehaving or injuring others will result in at most a slap on the wrist." Personal responsibility for one's own actions has sadly gone out of fashion.

In addition, superficially well-meaning authority has e.g. laid down that children must participate in making their own rules, rather than following fixed rules mandated by adults. The net result is commonly the total absence of set limits, coupled with the young's total disregard for adult admonitions and the adult's disinclination to get involved in conflicts concerning the children of others, and sometimes even their own.

The Swedish take on this has for many years been the deep conviction by authority that any problem, no matter how great, can be solved by simply providing "more information" to the public and parties concerned. Whenever some situation gets really out of hand, there follows a period of soul-searching (and costly investigation) about where the information flow went wrong. Because of course, the root assumption is that the people who did wrong, did so because they were not properly informed. (The Swiss take is somewhat similar, except that they place the responsibility of keeping adequately informed squarely on the shoulders of the individual.) The publicly repentant authority in the end sighingly assumes the responsibility (in the abstract, of course), vowing to inform better.

Really of course, I believe the problem is at root more to do with individuals opting out of personal involvement -- involvement with their family members, involvement with their neighbors, involvement with their schools, their community... -- involvement with the common values of the society in which they live. It is today all too easy to push this away, saying that authority XX has the responsibility to see to it that YY does or does not happen.

Abdicate personal responsibility and society soon degenerates and becomes progressively more dangerous, raising popular calls for more authority. The feedback in this process can rapidly lead to some really nasty societies to live in.

/ Bo

--

"Bo Leuf" bo@leuf.com
Leuf fc3 Consultancy
http://www.leuf.com/

Agreed. But I don't think it's so much a matter of individuals opting out as being forced out, at least here in the US. I'm 45 years old. When I was a kid, children who misbehaved had to keep an eye out not only for their own parents, but for any adult. There was an unspoken social contract. Adults were responsible for supervising children, their own and anyone else's if they happened to be the only adult present. An adult intervened when he saw any child misbehaving or in danger. And parents did not take exception to another adult exercising his best judgement to control their children's actions. Adults were presumed to have mature judgement and good sense and children were presumed to need supervision and discipline. And that system worked pretty well for the last few thousand years or so.

Nowadays, in this country at least, any adult who is foolish enough to intervene with a child not his own is liable to be sued or arrested. For that matter, simply because it is your own child is no guarantee that some interfering government bureaucrat won't have you arrested or put through the mill of "social services." Child abusers and child molesters are no more common nowadays than they've ever been, and arguably are less so. But the Politically Correct see child abuse behind every corner.

In Common Law, children have always been presumed incompetent to testify in court, and for good reason. But nowadays, adults may be jailed on trumped up charges based on perjured testimony by children, who are not even subject to cross-examination. Read up on the Little Rascals' Daycare Center scandal that happened near here several years ago. The owner, his wife, and all the adult staff were accused of sexually molesting children. When that story broke, I told my wife I thought it was ridiculous. Could this guy be a child molester? Certainly. But how likely was it that his wife and the entire staff were also child molesters? About zero probability.

It all got started because of a lie told by one child. The interfering government social services dorks started interviewing other children. Most children are eager to please adults, so naturally the social services morons got the answers they wanted to hear. So the owner of the daycare center and his staff had their lives ruined. Dealing with other people's children nowadays is very dangerous, as these people found out.

Any adult, particularly a man, has to be incredibly careful around children. Our good friends, Steve and Suzy Tucker, have two kids. Katie is a pre-teen, and Andrew is in elementary school. They're good kids, too. On occasion, Steve and Suzy have asked us to babysit. The Tuckers, adults and children, are the last people on earth I'd expect to hurt us. We've known Steve and Suzy for ten years, Katie since she was two, and Andrew since he was born. And yet, I make very sure that Barbara is with me all the time when I'm babysitting for Steve and Suzy. I can't afford to do otherwise. And that's a sad commentary on where the government and Political Correctness has taken us.

* * * * *

This from Chuck Waggoner [waggoner@gis.net]:

Your assessment of the high school tragedy furnishes much food for thought. It is bothering that issues of this era have become so emotionally felt, that intelligent investigation of cause and effect is frequently and irrationally brushed aside with calls for Uncle Sam to ban more of this or that.

When my son was about 7 years-old, I recall having to explain that there is a difference between passing a law to prohibit an offense, and actually stopping the offense. At that time, he thought that if there were a law prohibiting, let's say, the robbing of a bank, that no banks would ever again be held-up.

It is truly scary how many adults think that by merely passing more laws prohibiting every possible element of such a tragedy, it therefore would be wiped from the realm of possibility.

I've heard calls in the media for metal detectors to be placed in all high schools across the country, just as they are now in all airports. But, in this case, it appears the perpetrators stormed the school from outside. I doubt that metal detectors would have deterred these--obviously committed-to-the-death--individuals from their apparently well-planned spree.

My perception is that one huge problem our nation faces, is the concept that bigger is better. Smaller, once self-reliant communities have been gobbled up by massive city-county governments, and neighborhood schools have been abandoned in favor of huge county corporate ones that can't possibly tend to children in the more closely supervised ways, and with as much intimate knowledge of the kids and their families, that smaller school districts once did.

And unfortunately, I don't look for that to change anytime soon.

--Chuck Waggoner [waggoner@gis.net]

Exactly. And each new law does nothing to address the problem it was intended to solve, but results in a diminution of personal freedom. As Bo Leuf pointed out, calls for the government to "do something" eventually result in a society where none of us will want to live. The "cure" is much worse than the disease. I would love to see just one government spokesman stand up and say something like, "I'm sorry, but there is nothing we can do. There is no law we can pass or action we can take that will prevent things like this from happening. We've done too much already. Some of what we've done while trying to solve the problem has actually made the problem worse." But I won't hold my breath waiting for that statement to be made.

 


TTG Home

Robert Home

Daynotes Home

Search TTG

Special Reports

Last Week

Next Week

Monday

Tuesday

Wednesday

Thursday

Friday

Saturday

Sunday

Top


Friday, April 23, 1999

Sorry for the late update. I wasn't feeling well yesterday. I finally took my temperature around noon and found it was 100.4F (38C). Barbara says there's something going around, so it looks like I got it. I'm feeling better today, but I was a little slow getting started this morning.

It's time for my usual weekly network backup. This time, instead of using my Travan drive, I'm going to use the OnStream DI30 tape drive I've been testing. You can take that decision as a qualified endorsement of this drive. I don't take chances with my data. I've been experimenting with this drive for quite some time now, and, except for some bugs in the bundled backup software, I like what I see.

If you're thinking about buying a tape drive and had about settled on a Travan TR4/NS8 or NS20 drive, check the DI30 out first. This $250 ATAPI tape drive stores 15 GB natively (30 GB compressed) on $33 tapes, and provides 2 MB/sec native throughput. The drive has only been shipping for a couple of months, so some teething problems are to be expected. The only problems I've experienced with the DI30 are related to the bundled Echo backup software, which is still in its original "dot-oh" version.

OnStream tells me that they'll be releasing an updated version of the Echo software in the next month or so, and that it will incorporate many fixes for known issues. I expect that update will fix most or all of the problems I've encountered during exhaustive testing of this drive. I'm on the short list to receive a copy of the updated software as soon as it ships. Once I get it, I'll wring it out and publish what I learn.

At this point, the only reservation I have about the DI30 is the instability the Echo software exhibits when running under Windows NT, a problem that the next release of Echo should fix. Accordingly, I'm doing my production backup on a machine running Windows 98. I should have my review notes posted sometime next week.

* * * * *

Speaking of review products, Intel just sent me one of their new "Sun River" SR440BX motherboards and a Slot 1 Celeron/433 for testing, along with some other stuff I'm not allowed to talk about yet. The Sun River ($175 street) is essentially an upgraded microATX version of the ATX RC440BX I've recommended in the past. Like the RC440BX, the SR440BX supports Slot 1 versions of the Celeron, Pentium II, and Pentium III, and has embedded Creative 1373 sound.

The main difference (other than the form factor and number of slots) is the embedded video. The RC440BX provides 64-bit nVIDIA RIVA 128ZX graphics with 8 MB of SDRAM, which is certainly nothing to sneeze at. The SR440BX instead provides 128-bit nVIDIA RIVA TNT graphics with 16 MB SDRAM. Hard-core gamers may dismiss the original TNT in favor of video cards based on the current high-end nVIDIA TNT2 and 3dfx Voodoo3 chipsets. But the nVIDIA RIVA TNT was state of the art until recently is certainly more than Good Enough for the rest of us, and is likely to remain so for quite some time.

Writing that got me thinking about motherboards in general. I've used a lot of brands of motherboards over the years, everything from the major name brands to Pacific Rim garbage for which I could not identify the manufacturer. For everything that I think is really important in a motherboard--construction quality, stability, quality of the support web site, and so on--I've pretty much settled on motherboards made by Intel and EPoX. I consider motherboards from these two manufacturers to be the best of the best, so it's ironic that, at least among the major manufacturers, Intel and EPoX motherboards are often the hardest ones to locate when you want to buy a motherboard. They're worth looking for, however.

* * * * *

This from Bo Leuf [bo@leuf.com]:

Yes, it's been much the same throughout the western world. For a few years there is a veritable witchhunt for molesters under every rock and behind every door. With some very ugly and tragic situations for the people directly affected by the zeal of certain social workers. The paranoia never really lifts after such an episode, and lingering suspicions are easily aroused by perceived odd behavior or remarks by children.

Many children will today totally ignore anything adults tell them. In fact some younger hardcore cases will actively threaten to accuse the adult in question of abuse or molestation.

/ Bo

--

"Bo Leuf" bo@leuf.com
Leuf fc3 Consultancy
http://www.leuf.com/

My own post got me to thinking again about this case, so I searched the Internet for information about it. It was worse than I'd remembered, one of the most hideous miscarriages of justice I've ever seen. There's more information about the Little Rascals case here and here. The prosecutors and the social services people are the ones who should have gone to jail in this case.

 


TTG Home

Robert Home

Daynotes Home

Search TTG

Special Reports

Last Week

Next Week

Monday

Tuesday

Wednesday

Thursday

Friday

Saturday

Sunday

Top


Saturday, April 24, 1999

I hate Microsoft. I've probably mentioned that before. And I do, despite the occasional message I receive from someone who accuses me of being a Microsoft lackey. It's true that I don't hate Microsoft all the time. In fact, I usually rather like them. But their software does drive me mad all too often.

This time, it started when I was trying to do my network backup. I used the new OnStream DI30 to do a full network backup, but then I decided that it wouldn't be a bad idea to duplicate that backup with my regular tape drive. That's when it all started. That drive is on the resource server bastet, and bastet has been acting weird lately. Or perhaps I should say that the whole network has been acting weird. That started a week or so ago, when I was suddenly no longer able to access shares on bastet from other computers on the network. For some reason, bastet was no longer visible to the rest of the network.

The error message said "\\Bastet is not accessible. The network path was not found." Well, bastet showed up in Server Manager in the TTGNET domain, but I was also unable to access it from there. I decided to remove the computer account from the domain. Server Manager allowed me to do that, although the machine remained visible in the list of members of the domain. Attempting to delete it again displayed the usual message saying that although bastet was still listed, it in fact had been deleted and would be removed during the next automatic update, which should occur within the next 15 minutes. Well, I waited a lot more than 15 minutes--two days in one case--and bastet was still not deleted from the domain.

I've since figured out what's going on, aided somewhat by an error message that would be meaningless to most people. That message mentions a duplicate name on the network. Unless you know quite a bit about NetBIOS-over-TCP/IP, that message won't help much. As it was, that message was enough of a hint to tell me to fire up WINS Manager and check the NetBT mappings. Sure enough, bastet was bound to both the Ethernet adapter and to the NDIS WAN wrapper used by the dialup connection to my ISP. That accounts for the "duplicate name" error message, which refers to the fact that NetBIOS is confused. Unfortunately, the WAN wrapper is grabbing the NetBIOS traffic before the Ethernet card sees it, and nothing I do can change that.

I've tried everything. I've changed bindings. I've disabled every NetBT resolution method except broadcast on all machines. I've created HOSTS files manually on each machine to map IP addresses to host names. I've done the same with LMHOSTS. Nothing works. Nothing.

I know what happened. When I made bastet my WinGate server, I had to create a dialup connectoid for my BellSouth ISP account. When NT prompted me for a user name and password, I very carefully cleared the Domain field, knowing what a disaster it is to leave that field at its default value.

So here I am, nearly three hours wasted, and I still can't run my backup from bastet. I write books about networking, Windows NT, and the TCP/IP protocol. If I can't make this work, what hope would most people have? The real problem is that Microsoft Networking is built on the shaky foundation of NetBIOS-over-TCP/IP.

In NT4, all of the core services depend on NetBT. For example, when you use SMB (NT's core protocol) to access shared files on a server, you're using NetBT, whether you know it or not. That's regardless of the transport protocol you've installed. Even if your transport is purely TCP/IP or IPX/SPX, you're using NetBT to perform core network functions. Windows Networking, as implemented in NT4, is basically a pathetic peer-to-peer networking foundation with client/server functionality grafted on. Even NT5, although it can reasonably claim to use native TCP/IP transport, in fact still depends on NetBT for some core functions.

And don't even get me started on the bastardized Windows NT "domain" concept. Once again, Microsoft shows its peer-to-peer networking roots with domains, which are basically just workgroups on steroids. Novell implemented a true directory service when they shipped NetWare 4.0, and that's been years ago. Microsoft still doesn't have a directory service, despite their Orwellian attempts to represent the Windows NT 4 master domain directory database as a directory service. It ain't.

What's worse is that the vaunted Active Directory in Windows NT5 is arguably not a true directory service, but is instead simply a directory service-like view of an underlying flat domain structure. A true directory service is hierarchical in much the same way that a filesystem is. For example, in a hierarchical filesystem, you can have only one file of the same name in a given directory, but you can create a file of the same name in another directory. That's because the true (fully qualified) filename comprises both the actual filename and its location in the hierarchical structure. For example, the filenames \autoexec.bat and \dos\autoexcec.bat can coexist on one volume because their true filenames are different by virtue of the differing paths.

In a hierarchical directory service, the equivalent of a folder is called a container, and the equivalent of a file is called a leaf. If I organize my directory structure organizationally, for example, I might have one organizational unit container named SALES and another named ADMIN. If each of those departments has an employee named SMITH, I can create an account for each of them in the appropriate container. Although each account is named SMITH, the true names are actually SALES\SMITH and ADMIN\SMITH, two clearly different names.

I was shocked when I received the first beta of Windows NT 5 to find that I could not create identically named users in different containers. If I created user account SMITH in SALES and then attempted to create another account named SMITH in ADMIN, Windows NT 5 told me that I was attempting to use a duplicate username. This told me immediately that I wasn't working with a true directory service. I was working with an underlying flat domain structure that the administration utilities simply made look like a hierarchical directory. Perhaps this has changed with later beta releases, but somehow I don't think so. Novell will have a field day. Microsoft must be worried sick. If they aren't, they should be.

And if after all this you still think I'm a Microsoft lackey, please direct your comments to /dev/null.

 


TTG Home

Robert Home

Daynotes Home

Search TTG

Special Reports

Last Week

Next Week

Monday

Tuesday

Wednesday

Thursday

Friday

Saturday

Sunday

Top


Sunday, April 25, 1999

I confess that I still use dial-up Internet access. ADSL and cable modems are not yet available here in Winston-Salem, and ISDN seemed very expensive for the limited increase in throughput it provided. As a matter of fact, I've never even felt it necessary to upgrade to a 56K modem. My Internet access modem until recently was a USR Courier V.34+ modem that my friend Steve Tucker loaned me long ago. It was connected to sherlock, the WinGate server, through which it provided access to the rest of the network. The Courier normally connects at 31.2 Kbps, and that's been adequate for what we do here. Until very recently, 56K modems didn't do all that much better than 31.2 Kbps in real-word use. We're at the end of a long loop, and I suspected that 31.2 Kbps, if not the absolute limit for our line, was getting pretty close to it. So I skipped the 56K generation entirely.

And that brings up an interesting story. As a part of re-jiggering my network, I turned bastet into a resource server. One of the things I wanted it to do was run the WinGate proxy server software to serve as a shared Internet gateway for the rest of the network. Being a belt and suspenders kind of guy, I decided to bring up bastet as a WinGate Pro 3.0 server before I took down the WinGate Pro 2.0 server on sherlock. That meant I had to come up with another modem. I had a USR Sportster 28,800 fax modem sitting on the shelf, so that's what I used.

Running WinGate, the modem and the Internet connection is pretty much transparent to users. The software and modem are running on a different machine, you don't hear the modem dial, Outlook sends and receives mail in the background, etc. So it was several days before I began to suspect that all was not as it should be. When I started checking Dial-Up Monitor on bastet, I noticed that the Sportster was typically connecting at 24,000 or less, with a 26,400 connection perhaps one time in ten, and a 28,800 connection almost never. What was worse was that I was sporadically losing IP connectivity. The connection stayed up, but nobody was home. Dropping the connection and re-establishing it solved the problem until the next time, which might be anything from a few minutes to many hours.

At first, I attributed this to problems at BellSouth.net. They've been known to have such problems before. But after this went on for several days, I finally decided to swap modems. I pulled the Courier off sherlock (where it was still set up to receive faxes) and moved it to bastet. I put the Sportster on sherlock as the fax modem. Now everything works just fine. I'm back to connecting at 31.2 Kbps for the Internet link, IP connectivity doesn't disappear mysteriously, and faxes work just fine.

The problem may have been the modem configurations. The Courier and the Sportster are close, but not identical. The modem on sherlock was obviously configured as a Courier. The modem on bastet was also configured as a Courier, so this may be a case of the Sportster "almost working" with the Courier drivers. So perhaps it's not fair to blame the Sportster.

But I've had experiences in the past with properly configured Sportsters and Couriers that tell me that the Courier is the best modem there is for punching through and making a connection. Don't get me wrong. The Sportster is a very good modem. In fact, it's probably the second-best modem there is, after the Courier. But when you absolutely, positively have to make a dial-up connection, the Courier is the modem to bet on.

This has always been true. Back in the days when Hayes dominated modems, there was one niche where they had little presence. Bulletin Board System (BBS) Sysops and serious BBS users used US Robotics almost exclusively. This was partially because USR was smart enough to sell modems at a big discount to BBS sysops and partially because USR's proprietary HST protocol was faster than any standard protocol of the time. But the real reason was that USR modems made the connection when no other brand would. The first USR modem I ever used was a 300 bps Passport, back in 1981. Since then, I've used a lot of USRs, both personally and in a corporate datacomm environment, and I've never had a bad experience with them. If I needed to buy a modem today, it'd be a USR Courier. But what I'm using now is Good Enough, and I suppose I can wait until cable modems or ADSL arrives here.

Speaking of which, Steve Tucker tells me that he's been talking with a friend of his who's a tech with Time-Warner cable here in Winston-Salem. Apparently, they're testing cable modems now, and plan to begin rolling them out this summer. I hope so. One thing does worry me, though. Steve mentioned that he'd asked the Time-Warner tech if they'd be providing static IP addresses. The guy didn't know what he was talking about. That's one problem that a lot of people with cable modem service experience. Cable people are generally clueless about datacomm in general and TCP/IP in particular. My friend John Mikol has had a cable modem for at least a couple of years. For months now, he's been able either to watch cable TV or to access the Internet, but not both at the same time.

The other thing, of course, is service levels. Cable TV folks are used to maintaining reasonable up-time statistics, but nothing like what telephony and datacomm folks expect. Every month or two, we have cable outages that last from several minutes to perhaps a couple of hours. Following a bad ice storm it's not uncommon for some people to lose cable service for several days or more. That's not acceptable for something as important to me as Internet access. I may end up keeping my dial-up service even after we get cable modem service installed. Alternatively, I may see if any of those free deals are still available, like the old AT&T WorldNet deal that gave you five free hours a month.

I just hope Time-Warner is reasonable about home networks. BellSouth's policy on ADSL is that you're allowed to connect one PC to your ADSL line. If they find that you've connected a home network to it, you're in violation of your service agreement and they terminate service. Geez.

* * * * *

This from Rick Boatright [boatright@vocshop.com]:

Perhaps it's time to explore Novell Directory Services for NT. NDS for NT _does_ give you true heirarchical directory services... I know I know, but it's Novell after all... But still, perhaps it's time for another look. Netware 5 is a _loverly_ OS if you want to do SERVER stuff on the server, unlike NT which is a real peer to peer workstation os on steroids.

Rick

NDS for NT, hell. I may just start running NetWare, period. Actually, I'm a NetWare guy from way back. I never worked with NetWare 1, but I do remember installing NetWare 2.0 for clients back in the 80's. I was one of the first half dozen Enterprise CNEs in North Carolina, and also, I believe, the first Master CNE in North Carolina.

So why did I abandon NetWare for NT? Because NetWare people don't buy many books, and NT looked to become very hot. So, back in 1995, I immersed myself in NT and started writing books about it. But the fact remains that someone who wants rock-solid file and print sharing is better off with NetWare. Granted, as an application server, a NetWare box makes a good file server, but NT's attempt to be all things to all people means that it doesn't do any of them as well as it might.

Most people probably don't remember UnixWare, but in my opinion it was one of Novell's major strategic blunders to let UnixWare slip through their fingers. If Novell had taken a two-pronged approach--NetWare for file and print services with UnixWare as their application and Internet server platform, integrating both with NDS--NT would never have gotten off the ground. Instead, Novell let UnixWare wither on the vine and continued to sell NetWare as an application/Internet server platform. NetWare 3 was a horrible app server platform, and NetWare 4/5 isn't all that much better. Most of the NT boxes that sneaked in the back door at NetWare shops were used as application servers. They'd never have made it in the door if Novell had emphasized UnixWare as they should have.

 

 

Copyright 1998, 1999, 2000, 2001, 2002, 2003, 2004 by Robert Bruce Thompson. All Rights Reserved.