12 June 2012

My First Date with DAT

A piece I wrote for Which Computer -- certainly not the best review I've ever written. It falls into the common error of getting too bogged down in reporting problems the reviewer encountered, without giving a proper overview of what the device is supposed to do.


I was overawed by DAT. The idea of being able to pile that much data (a gigabyte was a huge amount back in the late '80s) onto a tiny, cheap cartridge became an obsession for several years. I followed DAT through its many iterations over the following decade, and still have an HP 2GB DAT device that somehow got left behind after a review. And a lot of old DAT tapes containing, I'd imagine, fascinating historical data. I have to imagine the fascination, because without a suitable SCSI controller card and a copy of whatever proprietary software I used to make the backup, I've no prospect of ever seeing that data again.
[The GigaTape 1230][wc?][13 July 89][chb]

The first company with a practical DAT system on the open market is the German firm GigaTape, known in the US as GigaTrend. Its GigaTape 1230 is a stand-alone cream coloured box with a foot print of approximately 9 inches square, standing a little under five inches high. Versions are available for the Macintosh and IBM PC environment: the review system hooked into our CompuAdd 386-25 by means of one and a half metres of stout cable that plugged into an AT SCSI adaptor board supplied with the system. This is a full length 16-bit card built by Western Digital, using high reliability surface mount technology.

Those used to the whirr and clatter of a traditional tape backup system will find the GigaTape 1230 a mysterious box to work with. The DAT cartridge is fed in through a small letterbox above the front panel, whereupon it disappears from sight behind a flap, spirited away by an actuator. That's the last you will see or hear of the tape until you eject it again. The machine is virtually silent in use, the only indication of action being the computer screen and a four digit LCD in the front panel of GigaTape device that racks up the units of tape travel.

Beside the LCD is a row of four pressure buttons marked OnLine, Unload, Setup and Test, whose plainly labelled functions belie the fact that you will have to work in very close conjunction with the manual is you want to do anything useful with them. You will also need to turn to the documentation to discover why the button beside the cassette slot marked Eject doesn't give you your tape back. "The button marked EJECT is not operative," the manual explains helpfully.

On booting up the host computer, the BIOS in the SCSI adaptor reports "no devices are responding" -- in fact correctly, as the SCSI interface is only activated later when you load the backup software. Unfortunately there is nothing in the very skimpy photocopied A4 manual to reassure you that this "error" is correct behavior.

The GigaTape device is scheduled to appear outside the PC world bundled with a variety of backup software. But it looks as if PC users are stuck with GigaTape's own LANSAFE software, developed in house in Germany. The first version of this we received was a mess of mixed languages, but version 1.04, arriving a week before going to press, had corrected most of these problems.

LANSafe works with DOS drives and can also backup and restore Novell devices, although it wasn't able to recognise NETBIOS type device names on our Invisbile Network LAN. It is a powerful, well-structured package, but like so much backup software lacks the finesse of more widely used products, and at first glance gives the impression of an engineers' utility to which a pop-up menu front end has been attached as an afterthought. Error messages tend to be terse and are flashed up on the screen only momentarily, usually without explanation. For example, a quick glimpse of the message "Read Error" down in the left hand corner of the screen as you attempt to mount a tape might lead you to supect that something is seriously wrong with the machine, or with the quality of the media. In fact all this message meant when we encountered it was that we had forgotten to initialise the tape correctly.

GigaTape uses a format called DATA-DAT, promoted by Hitachi and others. Unlike H-P's DDS system (see Technology Report), DATA-DAT requires the tape to be pre-formatted. Normally this two-hour process is carried out by the manufacturer, but to ensure you're not trapped into a single source for your tape supply, the GigaTape system allows you to format your own tapes, either through the LANSafe software (which means your machine will be tied up for two hours) or directly through the front panel of the GigaTape device.

Two further steps are necessary before the tape can be used to store new data. First you have to create a volume, giving the tape a name by which the tape library, stored on your hard disk, will identify it in the future. Each volume is further divided into what GigaTape calls "chapters" within the volumes. A chapter corresponds to what other backup system manufacturers would call a "session", or a "dataset". By this means multiple sessions can be recorded on the same tape -- obviously on most systems the 1.2 gigabyte capacity would be wasted if you couldn't do this.

LanSafe's library is a database of all your DAT tapes, preserving the directory structure of the files you have backed up. You can restore single or multiple files by choosing them from the library, and you can inspect the database either tape by tape or file by file. This second approach assembles files of the same name from several sessions, together with their creation dates and sizes, so that you can keep track of multiple revisions.

Conclusions

The GigaTape 1230 is very fast, streaming data at a rate of over seven Mbytes a minute.

The relatively high purchase price of the device, 5,000 pounds including LANSafe software, is offset by the remarkable low cost of the media -- around
10 pounds or less per DAT cassette.

The very poorly produced software and hardware manuals are off-putting at this price, but the unit itself is handsomely produced and a joy to use.

NeXT Swims into View

(I'm pretty certain this piece is mine, although the tag in my database says "anon". If you know different, please give me a shout)

Information about NeXT was thin on the ground three years after Steve Jobs had left Apple to set up his new company. What we did know was that NeXT machines were imminent. What we didn't know -- at least I didn't know -- was how to spell the company name. Or (I note with some chagrin) the difference between UNIX the operating system and Mach the kernel.


But despite the waffle, I think I get the thrust of this piece for "Which Computer?" right. With the arrival of the 386 processor the PC is beginning to invade the traditional workstation space. During the course of the roll-out of NeXT over the following three years Jobs will discover that his real USP is not the hardware but the operating system. But by the time he reconfigures NeXT into a software-only company, he's too late.

Er, except that being Steve Jobs he's allowed a resurrection. But that's another story...

[NeXT Workstation][wc?][1 Jan 88][anon]

(year right -- date a guess)

Although it is not exactly rare for unannounced machines to make a bigger splash before launch than afterwards, the forthcoming Next workstations deserve attention despite the hype surrounding them. Not simply because they are the first products to emerge from Steve Jobs' Next Inc., the company he set up after being effectively ousted from Apple in 1985, but because they promise to bring workstation power down to personal computer price levels. And that promises users the ultimate in desktop computing power, without having to give up the benefits of truly personal computing in terms both of accessibility and price.

Indeed, the Next machines symbolise the convergence of 'traditional' personal computers - if anything less than 15 years old can be said to be traditional - and dedicated engineering workstations. The difference between these product categories used to be clear, since personal computers had 16-bit processors, very limited RAM and disk capacities, minimal communications facilities, and poor-quality graphics displays; on the other hand, workstations had 32-bit processors, vast quantities of extremely fast RAM and disk space, built-in networking to share even bigger disks between groups of users, and enormous graphics screens with resolutions of 1000 x 1000 or better. Now, machines like the 80386-based PCs and the Macintosh II can, when suitably configured, provide all the features that a 'traditional' workstation can.

And while workstations were always cost-no-object products, bought in small quantities for professionals who really needed their power and capacity, the 32-bit personal computers are in the mass-market mainstream and priced accordingly.

In such a market, the question of whether the Next machines are PCs or workstations is largely academic. According to sources close to the company, the low-end Next system comes with a 32-bit Motorola 68030 processor backed up with a 68882 maths co-processor, 4 megabytes of RAM, a single 3 1/2in 1.44Mb floppy drive, a 40 megabyte hard disk, built-in Ethernetnetwork and SCSI peripheral interfaces, an integral fax modem, and a 17in monochrome monitor capable of displaying 1280 x 960 dots in up to 256 shades of grey. So far, so not very unusual, since these specifications are not too far beyond a machine like the Macintosh II today, and match what Apple itself might launch as the Macintosh III in 1989. The price too is personal computer-like, at around $5000.

But where the Next system edges into workstation territory is in its operating software and in its adherence to industry standards that have been created and adhered to by workstation originators like Sun Microsystems and Apollo Computer. The operating system will be Unix - or, according to some reports, Stanford's Mach - with a windows-and-icons front-end that may or may not come from the UK's own Cambridge-based IXI, a Torch spin-off. The networking will use Sun's Networking File System (NFS) standard to allow transparent file-sharing between Next systems and NFS- compatible machines from many other vendors. And the display will use Adobe's Display PostScript to create its graphics images, so that what is seen on the screen will precisely match what is printed on a PostScript-compatible laser printer.

Like workstations, the Next systems make no compromises and do not attempt to be compatible with existing personal computers in the marketplace, apart from the obvious fact that they use the same fundamental electronic building blocks. Any software developer wanting to use the special features of the machines must write new software, or adapt existing software, to match the special features of the new systems, and Steve Jobs is attempting to attract software developers just like the Apple 'software evangelists' used to do in the early days of the Macintosh.

The problem, for Jobs and Next as well as for existing workstation vendors, is sorting out the channels through which their high-specification but non-PC-standard machines can be sold in large enough numbers to be profitable. As long as workstations cost £15,000 and up for a minimum configuration they could be sold like minicomputers, but at £5,000 and belowthey must be sold like IBM ATs to make money for the manufacturers. That means using dealers or value-added-resellers (VARs), and the reluctance of dealers and VARs to sell complex Unix systems, particularly complex Unix systems costing the same as high-end PCs, is legendary.

Workstations need more initial specification, more installation work thanks to the inherent networking, and a lot more after-sales support since there is no such thing as an off-the-shelf shrink-wrapped Unix application; all Unix programs must be tinkered with and re-compiled at least to make them work on a new hardware configuration.

To counter this problem, Jobs is primarily aiming the Next systems at the higher education market, where the number of prospective customers is manageably small but the numbers of machines per customer is high, and where the tightness of cash makes untried hardware attractive as long as it is cheap enough. In other words, Jobs is trying to sell the Next workstations in exactly the same way as the Macintosh was sold into Universities in 1984.

But in the long run, the workstation manufacturers have more of a problem than traditional personal computer manufacturers like Apple and IBM. The personal computer firms can up the technological ante at will, while maintaining software compatibility and adding enough computing power and high-end applications to chip away at the workstation user base. To compete, the workstation vendors have to cut their prices, cut down their system performance to meet those prices, and so try to battle the personal computer giants on their home ground of low prices and dealer distribution.

That is a high-risk strategy for companies like Sun, Apollo and Next to follow. But the blurring of the boundary between workstations and PCs - particularly when Sun, for example, produces an 80386-based machine that runs MS-DOS and uses the same basic hardware as PCs from Compaq, ALR, and now IBM itself - makes it a necessary evil if the workstation vendors are to beat off the competition from below in the 1990s.

20 March 2012

DNA's Tea #FAIL

Douglas Adams' posthumous publication "The Salmon of Doubt" was largely assembled from material found on his computer's hard drive. It's definitely recommended reading for anyone curious about the workings of this remarkable mind. He was a friend, and I miss him.

But there's one horrifically technically inaccurate piece in the collection that demands correction. I attempted to fix this in my Microscope column in March of 2003 on the eve of the invasion of Iraq.

[msc103: Trouble Brewing][mscop][06 Mar 2003][chb]

Up north where I was born, IT was of course not the proud but battered industry it is today, but a much cherished late afternoon dining occasion that slotted neatly in between dinner (which mealy-mouthed Southerners insist on calling lunch) and supper. The 'igh part of the name meant that the food would probably be hot -- welsh rarebit, perhaps, or poached egg on toast. And the tea part of course defined the accompanying beverage.

Coffee in those days was not unknown, but it was certainly undrinkable, having been traditionally prepared in a device called a percolator, which recycled boiling water ad infinitum over the grounds to extract every last ounce of flavour, caffeine and base bean dross. Because, dammit, there was a war on, or at least there had been a war when the coffee percolator had arrived in our house as a means of making the most of rationing. Post war we began using real coffee beans instead of roasted acorns, but the extra expense probably wasn't worth the difference in taste of the final product. But don't talk about the war. Not now.

Tea we did know about. I've steered the subject of conversation around to the cup that refreshes but does not inebriate because I intend to hijack this month's column to correct a misapprehension about the preparation of the beverage put about by an ex-colleague of mine. Douglas Adams isn't widely regarded as an authority on hot drinks, and certainly during my association with him in the Doctor Who days, liquid refreshment at room temperature in foaming pint glasses was more the focus of attention. But Douglas is in print (in "The Salmon of Doubt") with a recipe for tea preparation that I fear his many fans may take seriously.

In an essay simply entitled "Tea", dated May 1999, Douglas remonstrates against our Transatlantic cousins, who are puzzled by the British devotion to something "that never seems to them to be a very good drink". With the incisive insight that characterises his best writing Douglas responds that the cousins are hardly in a position to judge because they've never learnt to make it correctly. "The American habit of bringing a tea cup, a tea bag, and a pot of hot water to the table is merely the perfect way of making a thin, pale, watery cup of tea that nobody in their right mind would want to drink," he notes. Touche. A step in the right direction since the days of the Boston Harbour incident, perhaps, when the water was stone cold. But no cigar.

So Douglas puts them right. "The water has to be boilING (not boilED)", he writes, "when it hits the tealeaves." Precisely. But at this point the lecture goes right off the rails. First of all he advises the use of Earl Grey tea. This ghastly concoction is named after a merchant banker whose cargo of China tea arrived at the docks after a stormy sea voyage accidentally drenched in bergamot oil which had leaked from the adjacent hold. Rather than junk the resulting muck, the wiley Earl, in one of the great marketing coups of all time, foisted it on the public as a scented delicacy. Snobs and suckers are still lapping it up to this day, despise recent findings that bergomot oil is carcinogenic.

But there's worse to come. Douglas goes on to recommend: "When the kettle has boiled, pour a little of it into a teapot... and pour it out again...." Having warmed the pot and added the tea you: "Bring the kettle back to the boil, and then pour the boiling water as quickly as you can back into the pot."

Uh-uh. If the water has boiled and cooled, even by a fraction of a degree, chuck it away and start again. Reboiled water, in the words of my grandma, "loses its goodness". Her trick was to use some of the half-way boiled water to warm the pot. Modern tea aficionados may resort to microwaving the teapot briefly.

The final gaffe Douglas makes is to suggest putting the milk in first and then adding the tea. In a footnote he observes: "This is socially incorrect..." but this incorrectness has "nothing to do with reason, logic or physics". Wrong, Douglas. I admit that my habit of adding the milk last dates back to the days I hung around Sloane Square trying to better myself with the debutante daughters of the landed gentry, but there is clear physical and logical reason behind the process. When adding a small amount of milk to a larger quantity of tea it's much easier to judge by eye when the mixture is right.

So for our American cousins, an important message. Use Assam, not Earl Grey. Don't boil the kettle twice. Put the milk in last. If that works for you, get back to me for further more general advice about the trouble you have brewing. Oh, but I promised not to talk about the war.

<>