Week of 4/26/99

As I've mentioned previously, I read a lot of books. Over the last thirty years, I've probably averaged one a day. Nowadays, when I'm on deadline, that may drop to a couple or three books a week, but I try to make up for that when I take a day off by munching through several that day. I read a lot of "serious" non-fiction books, but I also read a lot of novels.

Sometimes, not often, I stumble across an author whom I think may be destined for greatness. I found one such author recently. Her name is Caroline Llewellyn, and she writes some of the best mysteries I've read in years. New novelists usually need to write several books before they master their craft. The highest praise a new author can normally expect is to have his book reviewed as a "good first novel." That means the reviewer is making allowances for the author's inexperience, but thinks that author shows promise. I sometimes go back and re-read all of the books written by one of my favorite authors, and the progression from early efforts to mastery is almost always obvious.

That progression is not evident with Ms. Llewellyn's books, or perhaps I should say that it's not yet apparent. I've been unable to locate a copy of her first book, but her second and third books are superbly crafted. If indeed she does improve, she will truly stand in the first rank of novelists. As it is, I think her early work compares favorably to that of some of the best current novelists.

As I’ve also mentioned previously, I collect ultra-modern first editions. Ms. Llewellyn is not prolific, which is perhaps part and parcel of the quality of her work. Only her latest novel, False Light, is still in print, and I plan to buy a copy or two for my collection. If you want to do the same, you can get that book from Amazon.com by clicking here. Her earlier novels are available only in the used book market, so I’m going to have to track down perfect copies of all her other first editions. I notice that some are already selling for considerably more than their original list price, which is a good sign.

* * * * *

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

Well, we just brought out in-house NW5 system up this weekend. Since our NW4 system was the result of a series of upgrades from netware 1 version 4.16 to Nw 2.2, to 3.0 to 3.1, to several more 3.x's to 4, to 4.1 to 4.11, we decided it was time to really clean house, and build a new NDS system from scratch. We did that, then installed NDS for NT on our in-house nt 4 application server, and let it synchronize with the NW5 server, and Viola!!! Treble music!

Of course, you're right, NW folks don't buy as many books. They don't NEED to. :-) But there is some reality in the thought that the surge of NT books on the market made NT get a lot of MINDSHARE and that lead to marketshare...

As for unixware, without Novell pushing folks to write applications for it, it languished. The _real_ reason NT has been so popular is that in "small" networks, it doesn't demand a dedcated server, you can continue to operate peer to peer with your NT server box as the "best workstation on the lan" and far to many people do.

Rick

NW is a damn near perfect file server, it does backups WAY better than NT ever did. Netware border manager is a FINE internet gateway, and works better than wingate etc have ever done,

I’m not sure that authors had a lot to do with the success of NT4. We tend simply to write the books that people want to buy. It was Novell’s hideously incompetent marketing that allowed NT to gain a foothold. Novell started with nearly complete market dominance, had a killer product in NetWare 4.1—much better than NT4 in most respects—and yet failed to get that message across to buyers. But now, having once been handed a pearl of immeasurable price by Novell’s failure to execute in the marketing realm, Microsoft seems determined to hand it right back with their failure to execute in the technical realm. Had Microsoft shipped a stable, full-featured NT5 by mid-1998, Novell would probably not have recovered. As things stand now, I believe it is the future of Windows NT that is in doubt.

Tuesday

Tuesday, April 27, 1999

I just found a new search engine called EZResults. Actually, it found me. I noticed going through my web log that spider.easyresults.com had sucked down 224 pages, which is the whole of my site. Half expecting it to be some email address sucking program, I hit http://easyresults.com and found that it was a new search engine, or new to me at least.

EZResults has some drawbacks—I can't find a way to search for a phrase, for example—but its database is very up-to-date, and that counts for a lot. The home page states "Other search engines take days or weeks to add new pages or change existing pages. EZResult allows you to add or update your web page instantly." I tested that claim and found it to be the literal truth.

Yesterday, I mentioned the author Caroline Llewellyn for the first time. I was pretty sure I had never used the word "llewellyn" on any of my pages, so I did a search using d:ttgnet.com llewellyn as the search string. The d:ttgnet.com part tells it to limit the search to pages within the ttgnet.com domain, and the word(s) that follow tell filter the results with an implicit AND linking the search words. The search engine returned zero hits, because it had not parsed my site since I added that page this morning.

So I went to their Add URL page and added this week's page. I then hit the back button in my browser and re-did the search. Sure enough, that page showed up in the search results. It couldn’t have been more than a couple of seconds between the time I added the page and when I re-ran the search, so "instantly" is a fair statement of how quickly EZResults adds and indexes pages.

The fact that manual updates are instant is all well and good, but it remains to be seen how frequently their spider indexes sites routinely. I keep hoping that some search engine someday will start using a spider that’s smart enough to keep track of how frequently sites are updated and schedule passes accordingly. So far, I don’t know of one. Perhaps EZResults will be the first.

EZResults also makes up for some of its search syntax weaknesses relative to other search engines by using word stemming intelligently. I found that simply by entering a phrase in normal order, omitting stop words like "the" and "and" allowed me to locate occurrences of that phrase without much problem. The size of the index is also reasonable. When I was using it last night, I seem to remember that it was listing its index size in the 34 million page range somewhere, although I don't remember where exactly. When I checked this morning, it was up to 35,158,486 pages indexed. The jury is still out on this new search engine, but it appears to be one to keep an eye on.

* * * * *

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

I know you are anxious to eliminate all media advertising and pay for everything, but--after years of working in advertising-supported television and on the fringes of newspaper publishing--it's hard for me to see that ever coming about.

Dow Jones' 'Wall Street Journal' has had one of the few pay sites that has attracted any significant money, and apparently it's not smooth sailing, as they are soon to start an additional free site--see Monday New York Times online article.

We're all accustomed to sour grapes coming from the competition, but I've been tracking this pay vs. free issue in my trade magazines for some time, and from my reading, the following quote in the article from a top CBS executive seems dead-on.

"They're in a bind," said Larry Kramer, the chief executive of CBS Marketwatch. "They understood when they made their product part of the paid tier that they were giving away the mass market. The advertising dollars from the mass market have come in a lot faster than subscription dollars.

"They've created yet another brand name," he added. "That's the opposite of the way you should go on the Web."

Newspapers learned long ago that there was more money in selling advertisements than they could get from readers. Public Television is facing the same dilemma, and I suspect they will eventually be permitted to run full-length commercials, although less frequently than commercial TV, as they definitely are not getting enough money from their viewers to sustain their faster rising overhead.

Intrusive as they are, and as upset as it makes me to pay upwards of $60 for software like Quicken which then makes you look at ads, I think we're in for a long future of this kind of thing. Some months ago, you indicated surprise that books haven't added ads. Surely it can't be long before that happens, too.

And it's going to be interesting to see what happens with the WSJ's pay site--a slow death, maybe?

--Chuck Waggoner [waggoner@gis.net]

Well, it's not so much that I want to eliminate all advertising as that I want to have a choice. I see advertising getting its grubby mitts on more stuff nowadays, rather than less. The Greater Greensboro Open is now the Great Greensboro Chrysler Classic, although that's at least better than the name when K-Mart was sponsoring it. We now have 3Com Stadium, for God's sake. Where will it end? With all of us walking around with ads tattooed on our foreheads?

PBS is already running real ads. I was shocked not long ago when I tuned into Mystery and found that Travellers' Insurance was running ads before and after the episode. I saw another one somewhere else on PBS, this one by an investment company. And it was the exact same ad they were running on commercial TV. I'll never contribute another cent to public television. The periodic begging was bad enough, but running actual commercials crosses the line.

The same is true for American Movie Classics. I couldn't believe it a couple of months ago when I started to watch a movie on AMC and it began with a commercial. So far, both PBS and AMC are running commercials only before and after the content. But that won't last long. Soon, they'll have commercial interruptions just like regular TV.

In the end, I don't really care. I don't need television. I'd be perfectly happy without it. The best television show or movie ever made pales in comparison to even a mediocre book. I'd much sooner spend a couple hours reading a book than waste it watching television or a movie.

Wednesday

Wednesday, April 28, 1999

The newspaper this morning says that hundreds of thousands of PCs world-wide were wiped out by the Chernobyl virus. Interestingly, it may be that the author of Melissa actually did a back-handed favor for a lot of people. By creating a relatively benign virus, he got a lot of people thinking about viruses and caused a lot of PCs to be scanned that wouldn't otherwise have been.

* * * * *

The Register reports this morning that the Pentium III/550 is a a very hot processor, literally. Apparently, several people who have samples and who are not honoring the May 16 embargo date for releasing information about the Pentium III/550 have reported that it runs very hot. That's not really surprising. For a year or more, informed opinion has been that 550 MHz is about the absolute limit for the 0.25 micron process that Intel uses for the Celeron, Pentium II, and Pentium III. Many have speculated that the Pentium III/550 would never have been produced except as a stopgap intended to bridge the unexpected delay in the Camino chipset and the newer generation Pentium III CPUs it will support. My guess is that Intel doesn't really plan to sell a lot of Pentium III/550 CPUs, but will simply introduce it as a placeholder in the MHz wars. The real workhorses will be the new generation Pentium III CPUs that will ship with the Camino.

* * * * *

I see that Anand is finally starting to appreciate just how good Intel motherboards are. He posted a review of the new Intel SR440BX "Sun River" motherboard which rates it highly. He does ding it for the absence of an AGP slot, and points out that the embedded TNT video can be upgraded subsequently only by adding a PCI video card, but that really misses the point. The Sun River is not an overclocker's or hobbyist's board. It is intended to provide the foundation for a decent mainstream PC with good sound and video on-board. For its intended purpose, the SR440BX is a superb motherboard.

* * * * *

In reading about the execution-style murder of popular British television personality Jill Dando, I was struck by one statement made by the New Scotland Yard spokesman. He said, "We are appealing to any one else who may have seen a white male, late thirties to forties, quite tall, maybe up to 5ft 11ins ..." The implication that a man of "up to 5ft '11ins" (180 cm) is considered "quite tall" in Britain surprised me.

A caucasian man of that age in this country would be considered average at 5'11" tall. I say "of that age" because age does have a lot to do with it. In 1860, President Abraham Lincoln was considered a giant at 6'4" (193 cm) tall. I don't have the figures immediately at hand, but I seem to recall that the average US caucasian male in 1860 was about 5'5" or 5'6" tall. Improvements in health care in general and nutrition in particular have caused that average to increase over the years. My father, born in 1923, was considered tall at 6'1" (185 cm). I was born in 1953. At about 6'4", I am considered tall, but by no means gigantic. In fact, when I was a teenager I played pickup basketball as a guard because I wasn't tall enough to play forward.

Although I don't know for certain, I would guess that children born during the depression years and later during the war years probably grew up noticeably shorter on average than those born when high quality food was available in adequate quantity. I do recall reading when I was very young that the average caucasian adult male in the US was 5'9" tall. That would probably have been in the very early 1960's. Later, in the late 1970's I recall reading that the height of the average caucasian adult male was by then 5'10 1/2" tall. Presumably, those differences in average height reflect the growth to adult height of children born during the immediate post-war period and the aging out of the older, shorter part of the sample during the twenty years or so that intervened.

Although I am recalling these figures purely from memory and may be off somewhat, the point is that I found it surprising that British men apparently are noticeably shorter than American men.

* * * * *

This from Blair McKay [75103.1537@compuserve.com]:

Please forgive an ignorant question; I've had very little experience with firearms.

On your web page, you state that there is no reason for a safety on a revolver. Why is that? Is it because a revolver's trigger is harder to squeeze, or is there some extra "step" that has to occur before the gun can be fired?

The purpose of a safety is to prevent the unintentional or accidental discharge of a firearm. Broadly speaking there are two types of revolver, called single-action and double-action. Actually, there have been a number of self-cocking revolvers made, but they are oddities. One could attend gun shows every week for years and never see one.

At any rate, the original type of revolver, called single-action, requires the shooter to cock the hammer manually before each shot. One does not cock the hammer until immediately before firing. Once the hammer is cocked, a relatively small pressure (typically 2 or 3 pounds, 1 to 1.5 kg) on the trigger discharges the weapon. A safety is useless on a single-action revolver, because the weapon is inherently safe when it is not cocked, and it is not designed to be kept in the cocked position for any length of time. It would make as much sense to design a safety for a hunting bow when in the drawn position.

Single-action revolvers are functionally obsolete with a few exceptions, although some are still being made. They belong to the Wild West era, and have been largely superceded by more modern designs. They are still made for two reasons. First, the design of a single-action revolver allows it to be made very strong, appropriate for extremely powerful loads like the .44 Magnum and up. Second, they remain popular among people whose hobby is recreating the Wild West.

Original single-action designs did have one serious defect. When uncocked, the hammer rested on the rear of the firing pin, and the front of the firing pin rested on or near the primer of the round in firing position. That meant that dropping the weapon or even banging it sharply could cause it to discharge unexpectedly. Wise shooters always carried such revolvers with the chamber under the firing pin empty, thereby converting their six-shooters into five-shooters. Many years ago, firearm manufacturers repaired this design flaw by modifying the hammer so that it rested against the frame rather than the firing pin and adding a transfer bar that was only raised into position when the trigger was pulled. If the transfer bar was not in position, no amount of pressure on the hammer could allow it to contact the firing pin.

For most practical applications, the second type of revolver, called double-action, prevails. It is so-called because it can work in either of two modes:

First, one can cock the hammer manually just as with a single-action revolver. That method is most suited to hunting, target shooting, and other applications where accuracy is paramount. When used in this mode, a double-action revolver is exactly analogous to a single-action revolver. Once cocked, discharging the weapon requires moderate pressure on the trigger, typically 2 or 3 pounds, and very short trigger travel. A safety is superfluous in this mode for the same reason as on a single-action revolver.

Second, one can fire a double-action revolver from an uncocked state simply by pulling the trigger, but doing this requires much longer trigger travel and much more pressure, often 12 or 15 pounds (5 to 7 kg). It is impossible to exert this much trigger travel and this much force accidentally. One must consciously pull the trigger to cause the weapon to discharge. Once again, the purpose of a safety is to prevent unintentional or accidental discharge, which is clearly impossible in this mode, making a safety superfluous. This mode is intended for when you have to shoot quickly and speed is more important than accuracy, i.e. for self-defense. When you need to defend yourself, the last thing you need is a safety getting in the way. Without extensive training few people can hit even a man-size target at across the room distances when firing in this mode, which is why a revolver is not the best choice of weapon for self-defense.

Thursday

Thursday, April 29, 1999

The FDIC is at it again with the "Know Your Customer" rules for banks. If you live in the US and are at all concerned about your privacy, please visit the Defend Your Privacy web site and add your name to the petition. I did. The petition asks for your name, email address, and postal address, which may concern some. It did me, the first time I visited the site early this year. But that was before I realized that this web site is run by the Libertarian Party. You may or may not agree with their politics, but they are certainly unlikely to sell your name to anyone.

* * * * *

In the good news department, I found the following while catching up on Jeff Cooper's Commentaries, Volume 7, Number 2:

A Middle Eastern terrorist, Khay Rahnajet, did not pay enough postage on a letter bomb. It came back with "Return to Sender" stamped on it. Being of the usual intellectual development of a terrorist, he proceeded to open the letter. Maybe he learned from that experience, but considering what he started out with, I doubt if he learned much.

(The foregoing information appeared in "Firearm News" from Stellenbosch, South Africa.)

* * * * *

And, speaking of Libertarian web sites, this from Dave Farquhar [farquhar@freewwweb.com]:

I got a piece of mail from a colleague today that at first surprised me. Seems the Libertarian party is running a web site called www.stopthewarnow.com. I don't feel so bad about having turned into a hippie now.

The more I thought about the libertarian ideals, the more sense the connection made. This war is unconstitutional, a waste of American lives, and a waste of tax dollars. None of that's incompatible with libertarian philosophy.

I thought you and the rest of your readership (at least those in the United States) might be interested.

This thing sounds way too much like the standard political endorsement/solicitation letter. I think I missed my calling...

I hadn't heard about that web site, but it doesn't surprise me. The Libertarian Party has used the web extensively and effectively since the web began. Not surprising, really, when you consider that the LP appeals first and foremost to rational people and that a good many of us techie-types fall into that category. Thanks for telling me about the site. I've already submitted my petition.

* * * * *

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

Is anyone in the US who is advocating tougher gun laws in the wake of the Denver massacre paying any attention to news from Great Britain?

In that country, which--for a long time--has had some of the toughest gun laws of any western democracy: popular BBC TV host Jill Dando was killed point blank by a hand gun; in Manchester two men with multiple firearms, including an AK47 shot passers-by, wounding five, during a high speed chase with police; another man in Manchester walked into a pub, shot three men, one of them five times, then the shooter escaped and is still at large; and a man who left a gas station in Hounslow (near Heathrow airport) without paying, ended up firing at pursuing police (and nearby civilians) with both a handgun and an assault rifle, later taking a hostage (fortunately no one was killed)--all of this in the space of one week!

What's more, in spite of their rigid gun controls, British police authorities say the gun used to kill Dando "was common and easy for criminals to obtain."

If you believe the anti-gun lobby, none of this should have happened in a country where guns are banned.

Yes, we've already had calls for tighter gun control laws, including one from that moron sitting in the Whitehouse. Although you didn't mention it, I saw in the morning paper that there has been another high-school shooting, this time in Alberta, Canada. Two shot, one of them dead, and again by a kid wearing a trenchcoat. This in a town of about 8,000 population. This news item only made the News Briefs section on page 3 of our paper. Apparently, such incidents are no longer worthy of page one treatment, or perhaps the body count was not high enough to rate page one in our paper. Canada, of course, already has ridiculously tight gun control laws, but they didn't do any good here.

Friday

Friday, April 30, 1999

I got two chapters back from Pournelle, cleaned up the formatting a bit, and got them off to O'Reilly yesterday. The official first two chapters of the book have now been submitted in first-draft form. Things are starting to move along, although I must confess I'd hoped to be further along by the end of April.

* * * * *

The Pentium III serial number thing is back in the news. A Canadian company, ZeroKnowledge Systems has posted a web page that illustrates how someone with malicious intent can turn on and read your Pentium III serial number. If you have a Pentium III system and want to see how this works, turn off the serial number and then hit the ZeroKnowledge web page. This link takes you to the introductory page. When you go to the next page, you're given the opportunity to run the demonstration. Your computer will lock up and require rebooting. When you restart it, it will display your Pentium III serial number, which has been stored in a cookie. Bizarrely, Symantec has already included this demonstration in their anti-virus signature files.

* * * * *

This from Dennis Loretz [loretzd@gte.net]:

I am looking to get into the NT arena. First thing I plan on doing is obtaining some background (technical) information on NT Workstation before making the leap. Any recommendations for a general book for installation and general management for home use? I have fiddled with an NT workstation at work before, but not enough to feel comfortable with it as I do Win95. Appreciate the help . . .

Dennis J. Loretz
Email: loretzd@gte.net
Web: http://home1.gte.net/loretzd/index.htm

Good question. I'm not really familiar with any Windows NT Workstation books, although there are a lot of them. If O'Reilly published a Windows NT Workstation book, I'd say it'd be the best bet, but they don't. Windows NT Workstation is really just Windows NT Server with some pieces missing. I got started with Windows NT Server, so I never had any problem using Windows NT Workstation, and never felt the need to read any books about it. I think you're unlikely to find an NTW book oriented to home users, simply because NTW is still perceived as a largely corporate OS. Many of the readers of my web site run Windows NT Workstation. Perhaps some of them have some good titles they can recommend.

Saturday

Saturday, May 1, 1999

Today is an administrative day. I have a bunch of stuff to do, none of it interesting, but all things that need done. At least that beats tomorrow, when I have to climb up on the roof and blow out the gutters. The combination of monsoon rains (4" in one day) and all our trees deciding to dump their seed pods at the same time made a real mess. I'd better get to work.

* * * * *

This from Paul Robichaux paul@robichaux.net:

Bizarrely, Symantec has already included this demonstration in their anti-virus signature files

No bizarrity involved-- Intel called them and asked them to do so. I saw the story somewhere yesterday (maybe on news.com, but I can't find the link in my cache right now).

I think you're unlikely to find an NTW book oriented to home users, simply because NTW is still perceived as a largely corporate OS.

Microsoft's been running print ads (the ones I saw were all in _US News & World Report_) positioning NTW as the better desktop alternative for small businesses-- the one I've seen features some veteranian [sic] who needs more power and stability than Win98 offers. Pretty interesting. Here we are 5 months from the alleged release date of W2K, and Microsoft's launching an ad campaign targeting W98 and trumpeting NT 4.0! I wonder if the objective is to get people off Win98 and onto NTW, or to keep people away from Linux and on NTW.

Cheers,

-Paul

--

Paul Robichaux | paul@robichaux.net | http://www.robichaux.net
Robichaux & Associates: programming, writing, teaching, consulting

I didn't consider it bizarre that Intel requested it. Just that Symantec agreed to it. The chances that Microsoft will actually ship NT5 in October are slim to none. They may ship something they call NT5, but it won't be what they've announced NT5 will be. And why would they ship their flagship operating system this fall when (a) everyone will be too preoccupied with Y2K to think about it, and (b) they absolutely, positively have to ship a product that's feature complete and stable. I'm still guessing we'll see the real NT5 no earlier than Q1/00, and my guess is it will be more like mid-2000. They may ship a preview edition early, but I can't see the wisdom in that.

* * * * *

This from Jerry Mah, who asks that his email address not be published:

I usually don't participate in these non-computer conversations, however a particular comment hit close to home. Alberta is a province, akin to a state. Taber is the city where the shooting occurred. I live near Taber, AB. "Ironically," Alanis Morisette was reported to have hung out in Taber this past summer.

Taber is a little town in southern Alberta, known for their corn. Taber's website has their population pegged at 7214. I doubt that they've experienced much in a population explosion. Their town is very low key on this entire subject, and perhaps have shown Americans a side of Canada that you are unfamiliar with. Canadians are not the sensationalists that our southern neighbors are. We don't make a lot of noise, when our government introduces legislation, we tend to grumble and then accept it.

Not that I'm 100% familiar with Canada's gun control laws, however the statement that we have one of the "...ridiculously tight gun control laws..." I would say is a pretty extreme statement. I still don't know what stops any law abiding citizen to own a weapon. In fact I could easily go out, obtain a license, reg. and weapon. One of the primary reasons that "Canadian" owners complained so vehemently is that they were forced to "register" their weapons. There were also some weapons that were banned as well, as far as I know they were fully automatics, or weapons that were easily convertable to fully automatics. However, I know several people that still purchase, and still own their weapons. In fact, since we're basically in farming country here, it's mostly people who live in the cities who don't own weapons. Now, I suppose my argument could be that if you don't want to register your weapon, you've got something to hide. However, I'm not here to defend my government's laws. I do make the point that I fail to see how they are "...ridiculously tight..."

Increasily I think that we should be looking at these individuals who have caused these atrocities. To say that it is gun control, or the lack of gun control that has caused this is not the answer. Where there is a will, there is a way.

I would rather my e-mail address not be posted, so that every gun activitist in the world mail bombs me for my statements.

My opinions are my own, and not of my employer.

If you're referring to my comment that the shooting occurred in "Alberta, Canada", I am fully aware that Alberta is a province. I didn't mention Taber because I'd forgotten the name of the town and the newspaper was gone by the time I wrote those remarks. The newspaper reported the size of the town as 8,000, which I didn't verify. I'm sorry to hear that you live so close to the scene. That must be terrible.

I am not personally familiar with Canadian gun control laws, and was commenting simply on the basis of what others have told me. I consider any law "ridiculously tight" that prohibits average people from owning a useful class of weapon. If you can walk into your local gun store and buy a Colt 1911 .45 automatic or a similar defensive handgun and carry that weapon for self defense, I'll withdraw my characterization of Canadian gun control laws.

I never said that gun control laws or the lack thereof caused these tragedies. I simply stated that disarming people leaves them defenseless and that the presence of even one person who was armed and able to defend himself might have done much to minimize the body count at Littleton.

I doubt you would have received much mail, if any, from pro-gun people. It's the hoplophobes who are vociferous when someone comments favorably about guns. It's kind of like the PC/Mac issue. I could post an article saying nice things about Macs and bad things about PCs and I'd likely receive few or no nastygrams from PC advocates. If I did the opposite, I'd be covered up with mail from Mac bigots.

* * * * *

This from someone who asks to remain anonymous:

I found your webpage on searching tape backups. I was wondering which model of HP uses the TR-4. The Colorado states that it uses TR-3.

My employer's office needs a backup system desparately. A local tech has suggested an HP 1559A DAT drive. This runs in the neighborhood of $800 for the hardware alone. For a 2 PC peer to peer with a 2.0 and 1.0GB harddrive this seems excessive. I am exploring the options. Either the above DAT, a Zip or Jazz, or Travan type tape drive.... I am trying to make the point to her that the DAT is overkill considering the fact that there is plenty of downtime when we don't need to access the system. 40mins to an hour is fully available to us. We don't need killer speed or bus mastering. It would be nice however to backup to one tape and not have to sit and guide the process, if you know what I mean. So on that count the Zip is probably out. The Jazz also moves into the neighborhood of $550 with a 2GB disk, I think. Of course, as you pointed out the tapes for Travan are $30 or so dollars. For fault tolerance, wouldn't you want at least 2 of these. I know that I can get a TR-3 drive for about $200.. So there you would be in about $260. My only worry is the tapes getting degraded. What are the optimal storage conditions? Then there is cleaning the drive. I know too well about that ... Our PC at home has a Conner backup drive that failed and some of the tapes were ruined by heat or whatever? Not sure.

Sorry for rambling.. Thanks in advance for any information you might give me.

You definitely shouldn't buy a TR3 or earlier drive. These are functionally obsolete. Neither should you depend upon ZIP, Jazz, or similar removable media. Tape is still king for backup, and for good reason. Unless there's something significant you haven't told me, I can't think why your tech would have recommended a DDS (DAT) tape drive. That sounds clearly inappropriate to me.

What you probably should buy is a Travan TR4 drive, assuming that 6 GB or so per tape is adequate to cover expected growth. Travan TR4 drives with IDE interfaces are available for less than $200 if you shop carefully. They store 4 GB natively, but are advertised as "8 GB" drives, on the assumption that your data will compress 2:1. I get more like 1.5:1, so that means about 6 GB will fit on a TR4 tape. TR4 drives advertised 30 MB/min throughput native and 60 MB/min compressed. My experience has been that they run at about 35 to 38 MB/min when backing up local data using compression on a reasonably fast machine, and do about 22 MB/min across a 10BaseT Ethernet network.

If you are running Windows 95, you may find that the bundled backup applet is insufficient to meet your needs. It does not include support for scheduling (although you can kind of get around that), for saving sets (you have to make your backup selections from scratch each time), or data compression. The backup applet that comes with Windows 98 is much better, and will probably suffice. In any event, make sure that the drive you choose is supported natively by the operating system you're using.

If you think you may be adding a machine or two to your network in the forseeable future, or if you may be replacing your disk drives with larger ones, you may outgrow the capacity of TR4 quickly. If that's the case, take a look at the OnStream DI30 drive. It costs about $250 on the street, uses $35 tapes that store 15 GB natively, and has about twice the throughput of the TR4 drives. However, it uses only its own backup software, called Echo, which may or may not be a problem for you. The OnStream drives are still very new, and the current version of Echo has some problems, particularly under Windows NT, but OnStream plans to release an updated version of Echo soon which should solve most or all of the problems I've had with this "dot-oh" release. I'm backing up my network to an OnStream DI30 as I write this.

Whichever drive you buy, plan to buy more than one or two tapes. In isolation, $30 or $35 per tape sounds expensive, but it's cheap insurance to protect your business. There are numerous tape rotation strategies, but all of the decent ones require at least six to ten tapes. I don't have room here to detail the issues, but any decent book that covers backup should at least provide the essentials. I think I did a rather decent chapter on backup in my book Windows NT Server 4.0 for NetWare Administrators. It may not be worth buying the book for that chapter, but you might want to see if your library has a copy.

Regarding cleaning, you are correct that it is critically important. My TR4 drive recommends cleaning after every ten hours of use using either a cleaning cartridge or a cotton swab with rubbing alcohol. I prefer the latter. When I first got the drive, I neglected to clean it as often as I should have, and I started getting errors on backups. Since I started cleaning it periodically, I've yet to get an error.

* * * * *

And this followup, which I've edited to remove quoted material from my original response:

Why so? Just curious. What makes his recommendation inappropriate? To me it seems far too expensive and performance oriented for the situation. We don't need to have bus mastering or high speed. But what of the tape for DATs? Are they reliable? How do they compare with TR4? I know that they are digital, of course, but also that they store data helixically (sp) for higher capacity on a smaller tape. I personally don't care how big the tape is. Is it more reliable, though?

What of Jazz drives? Are they reliable? They to be fairly compatible and dependable. Correct me if I am wrong, please. I don't know about Zip because that would necessitate a backup set rather than one clean store to disk. That seems a bit tedious to me, but then, no more tedious than backup to a floppy. A tedious and uncertain prospect at that. One that has gone on far too long.

The original drive you mentioned, at $800, is an inexpensive DDS drive. It is likely a DDS2 (4/8GB) drive, which will offer about the same amount of storage and the same performance as a TR4 drive. It will require installing SCSI, which is not inexpensive or simple. The only advantages of DDS2 is that the tapes are somewhat cheaper than TR4 tapes, that it may provide read-after-write (which allows you to do backup and verify in one pass), and that it may support hardware compression. But it's simply too expensive a technology to be reasonable for what you want to do, and the lower cost per tape is subsumed by the cost of the drive for your application.

Jazz drives are unworkable simply on media cost. They have a very high cost per megabyte relative to tape, and you'd need more than one Jazz disk to do a single backup. To support a rotation, you'd need even more.

Hmm I would say given the system she has it would take well over an hour to backup the 2GB harddrive. I am assigning your network transfer as her system is not very fast.

Why worry about how long the backup takes? Start the backup when you finish work and allow it to run overnight. When you come in in the morning, stick last night's tape in your purse and insert the next tape in the drive. If that's a problem because you currently turn your computers off at night, get in the habit of leaving them on. The computers will last longer, and the amount of electricity they consume is trivial, especially if you have the monitors blanked.

I think I could get around the scheduling part. Won't need compression or sets. But then there is other backup software that will allow the functions of scheduling, saving sets and compression, right? We run Windows 95 OSR2 BTW.

Yes, there are any number of other backup apps you could run. In fact, your tape drive may come with one. Just make sure that it allows you to back up network volumes that are mapped as drives on the local machine where the tape drive is installed. Most do allow this, but some "personal" backup applications will backup only local volumes.

I saw those OnStream units at CompUSA. Are they reliable? Sounds like given the relatively small harddrives needing backup that this might not be worth the possible headaches with such a new product.

Well, who knows if they're reliable? They've only been on the market for a couple of months now. The drives and tapes appear to be well constructed, and my guess is that they'll be very reliable. I've been beating up an OnStream DI30 internal IDE drive for a month or two now, and it seems to be a good technology. I have some qualms about the bundled Echo backup software, particularly under Windows NT, but those should be addressed by the new release, which OnStream tells me will happen soon. And I think it's a mistake to buy a backup solution based on your current hard disk sizes. What happens when you add one more machine, or when you want to install a new application or version upgrade and find that you need to install larger hard disks? Plan for where you're likely to be in a year or two, not where you are now.

Yeah, I know. A rotation is necessary. But how about a 3 tape rotation? Is that acceptable. Not that much changes on the system each day. I was thinking backup twice a week with rotation at most...

No, a three tape rotation is completely inadequate, unless all you want to guard against is catastrophic disk failure. Any decent backup rotation has as one goal allowing you to recover older versions of files. What happens if an important file is accidentally deleted or becomes corrupted and you don't notice immediately? You want to be able to go back a week or a month and retrieve an older stored version of the file.

Could you offer a recommendation for a TR4 drive? Seagate? Connor? HP? Colorado doesn't show itself to be a TR4. Also, what do you think of Sparq drives (sp)? My bosses husband has this for backup and he likes it. He also is very chauvanistic and will probably dissuade her from following any advice I give.

Connor is now owned by Seagate, and Colorado is a division of HP. I have a Seagate/Conner TapeStor TR4 drive, and I've been pleased with it. Other than that, I can't recommend specific brands because I haven't used them. Your boss's husband may be less pleased with his SparQ drive when he finds that SyQuest has gone out of business.

In any event, I am trying to learn about backups in the process. I need to present my employer with a solid recommendation for backup. In my gut, I don't think the tech's recommendation is right. Also, my boss is pretty tight-fisted. I think if I could present a cheaper, reliable option she would be happy to save the money. She bought the cheapest PC she could get.

She is a afterall a financial planner :).

Thanks in advance.

It sounds as though your boss tends to be penny-wise and pound-foolish. Many small businesses that fail do so because they've lost their data because they failed to keep a good set of backups. Ask you boss where she'd be if she came in one morning and found her PCs had been stolen, or that they'd been destroyed by fire, flood, or other catastrophe, or simply that the hard drive had failed. Does she have important data stored on them, such as customer files, billing data, etc.? Could she reconstruct it in a timely manner? Could she reconstruct it at all? A small business that does not establish a good backup plan and follow it is an accident waiting to happen.

* * * * *

This from Peter Thomas [peterjt@netcom.ca]:

I know at one point recently, you mentioned having heard of a 35mm "digital film" that would convert any existing SLR camera to digital.

Take a look at http://www.imagek.com

Thanks. Actually, I think it was Bo Leuf who made that reference. I'm the one who thinks such hybrids are a very bad idea.

I went over and checked out this page. I see that they're not actually shipping product yet, but plan to do so sometime this summer. It's an interesting idea, but for $800 I don't think it will fly. It's a question of too little functionality for too much money. The resolution is only 1280X1024, which is now mid-range. I didn't search the site exhaustively, but I saw no mention of CCD size. If they are in fact providing a full 24X36mm frame, I'd think the resolution would be much higher. My guess is that they may be using something less than full frame, which would turn a normal lens into a telephoto, and a telephoto into a super-telephoto.

But the real problems are caused by the fact that this device is a drop-in replacement for a 35 mm film cassette. The device has no on-board data compression, and stores only 24 images. When you reach that number, you must remove the device and transfer your images to a computer before you can take more pictures. It's not clear to me how long that will take, but if they're using serial transfer, it might take hours to download 24 uncompressed 1280X1024 images.

There's obviously no way to view images, or to delete ones you don't want to keep. Because the device is completely internal, even providing indications to the photographer is a problem. They mention audio prompts, and they also mention an indicator light, although that will not be visible with most 35mm bodies.

My guess is this device will fail miserably in the market. At $800, it's too expensive for casual users. Most of those who are willing to spend $800 on digital photography will buy a digital camera for its many advantages over this device--swappable memory sticks, a viewfinder that allows you to view and delete images, etc. Those who want to use their existing 35mm lenses are likely to find this device too constraining. I think it's doomed.

What I really want is a $500 to $1,000 digital camera body designed from the ground up to use 35mm lenses and accessories. If Nikon, Olympus, Pentax, Canon, and the other manufacturers of traditional 35mm equipment introduced a digital body as a part of their standard line, they'd sell a ton of them. But I don't think the technology is quite here yet. At a minimum, they'd need a full 24X36mm CCD with something like 2048X3072 resolution or better. They'd also need, at a minimum, image transfer via USB, if not IEEE-1394. Moore's Law makes me suspect that day is not all that far away.

* * * * *

Early Afternoon: I got most of my administrative stuff done, and now I'm juggling installations on a couple of test-bed machines. I think I'm going to do something I've never tried before: build a system without a case. I'm going to place the motherboard flat on my credenza and wire it up to a stand-alone power supply, floppy, CD, and hard disk. We'll see what happens. If FCC rules have any basis, it should create quite a bit of interference with other electronic devices nearby. I'm beginning to feel like Dr. Frankenstein. Now I need an Igor.

* * * * *

This from Bo Leuf:

Trust a reader to find the link for us.

However, you commented:

"The device has no on-board data compression, and stores only 24 images. When you reach that number, you must remove the device and transfer your images to a computer before you can take more pictures. It's not clear to me how long that will take, but if they're using serial transfer, it might take hours to download 24 uncompressed 1280X1024 images."

This is incorrect as far as their FAQ goes. 1. The device uses onboard lossless compression. 2. Transfer of the 24 images from device to pc is said to take *less than a minute*.

I'll be keeping an eye on this product. And future versions.

/ Bo

--

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

I'm sure you're right. As I said, I didn't spend a lot of time on the site. I still think these guys are going to get their butts kicked by the Canons and Nikons of the world. But to each his own.

Sunday

Sunday, May 2, 1999

I got the case-less testbed system built. It almost worked. It's a bit strange to start the computer by using a screwdriver to short out the two pins intended to connect to the power switch, but other than that everything worked fine. Or would have, except for the locked CPU multiplier. The SR440BX "Sun River" motherboard came jumpered for normal operation. When I fired it up, the boot screen told me that I had a 350 MHz Pentium III processor installed and then locked up. I shut it down and changed the jumper to the "configure" position.

When I restarted the system, the boot screen told me that I had a 200 MHz Pentium III processor installed. After a pause of several seconds, the system entered BIOS Setup mode. I attempted to set the correct CPU speed, but found that the highest available setting was only 450 MHz. I can't believe I just wrote "only 450 Mhz." How quickly we become jaded. At any rate, I set the CPU speed to 450 MHz, because I do have 450 MHz versions of the Pentium II and Pentium III CPUs around here somewhere.

People seldom think of it that way, but the CPU multiplier is as effective a bar to underclocking as it is to overclocking, so I wasn't able to run the CPU I'd intended to use. I checked the Intel web site and found that there is an updated BIOS available, although I was out of time by then.

* * * * *

At the same time I was doing all this, I was also working on rebuilding Testbed1. I used to give my testbed systems real names, but I tend to get too attached to systems that have proper names. The old joke says that lab technicians prefer to use lawyers instead of white rats because they get too attached to the rats, so perhaps I should name my testbed systems for famous lawyers.

At any rate, testbed1 is a Celeron/333 system built around an EPoX EP-BXT motherboard. It originally ran Windows NT4/SP4. I installed the OnStream DI30 tape drive in that system, intending to run it under Windows NT. What I found was that the dot-oh release of the OnStream Echo backup software didn't get along very well with Windows NT. That'll be fixed soon, when OnStream ships an updated Echo, but meanwhile I needed to test the drive. So I installed Windows 98 on that system. [As it turns out, OnStream Echo works fine under Win98].

I'd originally installed Windows NT in a 1 GB partition, but there was more than 3 GB of unpartitioned space on that drive. WinNT and Win9X are easy to set up as dual-boot if you install Win9X first. You simply create a primary partition for Win9X and devote the rest of the disk space to an extended partition. WinNT is perfectly happy to reside on a logical volume in the extended parition, although it installs its startup files in the Win9X primary partition.

But WinNT was already living in the first primary partition, so the easy answer was to create a second primary partition, mark it active, and install Win98 there. That meant that I had to switch between booting WinNT and Win98 by using fdisk to reset the active partition, but I could live with that for a while. It was, at least, a quick way to get Win98 running on that box so that I could test the OnStream tape drive. That was a temporary solution at best.

Rebuilding testbed1 properly meant I needed to have Win98 in the primary partition and WinNT installed on a logical volume. The easy way to do that was simply to blow away everything on the disk drive, use fdisk to take it down to bare metal, and reinstall everything. I got as far as getting Win98 installed yesterday, and should get WinNT installed and configured today.

Barbara is out working in the yard as I write this, and won't have time to clean house today. So I'd better get to work cleaning house and doing laundry. It's cool today, with winds gusting to 20 MPH (32 KPH) and a wind chill near freezing, so I don't think I'll climb up on the roof to blow out the gutters.

* * * * *

This from joshua [catpro@catpro.dragonfire.net]:

I don't consider it that bizarre for Symantec to consider the ZeroKnowledge program a virus. Intel is a big company, and they leaned on Symantec to do so.

What is bizarre is that Intel thinks that this is the proper way to deal with security holes in their chip. What if I use linux instead of MS (at the moment, although this is only temporary, the only computer of mine with Windows on it is a broken 486 notebook)? Now I'm exposed again to their flaws. Heck, it probably wouldn't even be that hard to modify the ZeroKnowledge hack to bypass Symantec. What will Intel do then? They can't block everything.

--

Joshua Boyd
http://catpro.dragonfire.net/joshua

I'm not particularly concerned about the CPU serial number as a security issue. What worries me more about it is that software vendors will use it to "marry" operating systems and applications to a particular processor. I think we'll see this start to happen with vertical-market applications real soon now, if it hasn't happened already. The bad old days of copy protection may soon return. Although no one has made much noise about it, I understand that the CPU serial number is not unique to the Pentium III, but is also included in Deschutes-based Pentium II CPUs and Mendocino-based Celerons.