Dan's Data in Retrospect

[LongRunner]

While growing up I read the hell out of Daniel Rutter's site, and I've recently downloaded it all for quick access and safekeeping.
Combining IT journalism with a touch of Australian resourcefulness, it was a memorable place.

Here's my assessment of the best (and worst) parts:

Technically Strongest
CPU cooler round-up; objective measurements of thermal resistance were more than even many major sites (including Hardware Secrets) did.
The only thing really missing was a general mention of the fan models on them.

Best Aged
The DIY UPS, a small diversion in 2001, is becoming ever-more attractive given the atrocious quality of newer UPS models (as lamented many times in Behemot's output waveform database). If I don't find an old good UPS to repair in the meantime…

HDTV on the cheap (from 2002, with a set-top box outputting to a VGA monitor) may come second, given the lousy quality of modern TVs (Mum's less‑than-a-year-old LG set already has visible burn-in and the remote markings are almost completely rubbed-off) and presence of HDMI input on most modern monitors (enabling direct connection of modern game consoles, etc.), on top of the generally increasing irrelevance of broadcast TV.

And of course mechanical keyboards have become much more popular than they were in 1999, and USB charging is now the norm.

Most Epic Product Fails
Pseudoscientific products aside, the Just Cooler FA-100 fan alarm was pretty much useless at its purpose, while the TherMagic CPU Cooling System was an all‑in‑one water cooler that performed no better than an average air cooler, while loud, bulky and overpriced (even if you disregard the manufacturer's recall). (The competing Aguatec cooler was less embarrassing, but no more useful.)

Low points
Obviously the PSU “reviews” weren't much good, but that was normal at the time.

Possibly less forgivable was him joining in the apologism for IBM's Deathstar 75GXP/40GV/60GXP drives, and IBM's retroactively-claimed runtime limit on them. (It's worth noting that from 2006, HGST gave an explicit 24×7 rating in Deskstar datasheets.) Even as a kid, I was skeptical; and sure enough I got almost 40,000 hours out of my trusty ST380011A, even after spending much of its life too hot to touch (67°C peak according to S.M.A.R.T.)
Probably it still would have made it to the “full” 43,830 hours, if not for suffering several months in a very humid shed after I stopped using it…

Especially since Maxtor promptly preyed on this FUD by deliberately scamming buyers with their DiamondMax Plus 9/MaXLine Plus II twins; made so they could survive 24×7, but willfully ground their heads and landing zone into oblivion after maybe 1,000 cycles give-or-take (you can easily hear how horribly rough their head take‑offs and landings were – whether compared to previous Maxtors, the concurrent and OK 5400rpm DiamondMax 16, or any other manufacturer; and the later 500GB, 4‑platter DiamondMax 11/MaXLine Pro was worse still – though to be fair, even the rare ramp‑loading DiamondMax 17 fell apart at times).
Eventually Apple and other furious OEMs forced Maxtor to fix their start/stop rating (on later DM10 and the last few DM+9 batches), but the wider public showed no mercy (their data having been already-obliterated beyond any recovery) and was glad to drive Maxtor out of the business as quickly as possible.
(It didn't help when enthusiasts bought the MaXLines, used them on/off and crashed them just as quickly as the regular DiamondCrashes – or indeed even quicker given their 3/4 platters instead of 1 or 2 in the smaller models – and Maxtor couldn't afford and refused to honor the 5‑year warranty )
The only further shame was how Maxtor's vicious managers then invaded Seagate after the buyout; after the already very-substandard but still vaguely-usable Pharaoh (7200.12/SV35.5/DM23), Seagate went the full-Maxtor with the infamous Grenada (ST3000DM001 and relatives) – big, fast, surprisingly power-efficient and quiet, and virtually guaranteed to head-crash by 3 years (despite now using load/unload technology); and no prizes for guessing what excuse Seagate tried to pull when people formed the 2016 class-action lawsuit (they weren't having it, but that didn't stop Seagate's equivocation from appeasing the judges).

After the 2016 lawsuit and reckoning concluded, Seagate was finally forced to make even Grenada half-decently under new model numbers (although it's still nowhere close to the classic Barracudas up to 7200.8, of course; that honor went to the Barracuda XT/“Constellation ES” line, and now to the FireCuda 3.5″).

But probably his most catastrophic error of all was here, where he claimed that the power filter was arranged to block DC instead of RF (and the circuit was so obvious from the photo! )

[Behemot]

In mention of UPSes, APCs Smart's are still pretty OK, if nothing less, than after refurbishing. Decent models though, stay away from SC, (SM)C and other with too many corners cut off.

IMO the best of all (so far) in my own experience are WDs RE4, 2TB, 7k2 models. Never actually seen any bad yet, got multiple of them in NAS (which is, due to electricity costs which I have to pay for a few years now - was not a concern when I did not have to really , now in hit-and-run mode, so I turn it on once in a while when I need to download something, upload new movies and series which I've seen since last time, and turn it off - they handle that all, despite being intended for more like all-the-time-on mode). Recently got two servers with 16 more of them for laughable money so I can put them into old SOHO NAS boxes because of all those good experiences.

I presume that being enterprise, Seagate Constellation and Hitachi Ultrastar are similarly good though.

[LongRunner]

IMO the best of all (so far) in my own experience are WDs RE4, 2TB, 7k2 models. Never actually seen any bad yet, got multiple of them in NAS (which is, due to electricity costs which I have to pay for a few years now - was not a concern when I did not have to really , now in hit-and-run mode, so I turn it on once in a while when I need to download something, upload new movies and series which I've seen since last time, and turn it off - they handle that all, despite being intended for more like all-the-time-on mode). Recently got two servers with 16 more of them for laughable money so I can put them into old SOHO NAS boxes because of all those good experiences.

The Caviar Blacks (which I've used) shared the same platform for high-capacity models, and came with a 5-year warranty too; I suspect they were WD's way of selling drives which didn't quite test up to the ultra-high “enterprise” standard, but were fine for general use (certainly mine survived well past 5 years, even 24×7).
Since they have head ramps (like all half-decent drives today), I wouldn't worry too much about a few more start/stop cycles (as long as they aren't overheated).
(Such “binning” in production has been common practice in one form or another even for 1980s drives, which were sorted based on their ability to handle the more‑efficient but less‑tolerant RLL (2,7) encoding, rather than the usual MFM as on floppy disks. Low‑capacity Caviar Blacks actually shared their platform with the Blues, but the process may have been similar – good drives sold as Blacks with RV sensors on the PCB, marginal drives as Blues with no RV sensors.)
EDIT: There are also single-platter RE/Se drives, which I guess made it a 3-level binning process altogether.

To be fair, though, I haven't personally had any decent desktop drive fail sooner than 43,830 hours under normal conditions.
(Seagate Barracuda 7200.11/ES.2 and 7200.12, let alone the 7200.14, were not decent in my experience…)

I suspect the mid-2000s HDDs (Maxtors notwithstanding ) were so robust in part because they had to survive both “consumer” (high temperatures, rough handling, frequent start/stops) and “enterprise” (long continuous operation, possibly higher I/O) stresses, not having (yet) been market-segmented.
(That said, it could well be hard to make current high-density drives as resilient – the already nanometers-thin protective coatings may be even thinner now… But at least we now have the option of helium-filled high-capacity HDDs, which are completely sealed against anything corrosive in the ambient air )
Seagate's U Series 5400rpm drives were truly consumer-grade – reliable enough when new, but much more prone to media degradation than the Barracuda ATAs, even though Barracudas are hotter 7200rpm drives (which was compounded by using linear voltage regulators up to the non‑NCQ 7200.7s, combined with the CBDS routine regularly scanning for errors during idle time) – but at least the U Series (especially U5, U6) were so slow that enthusiasts avoided them anyway.

Once the market segmentation was established, lower-capacity (particularly single-platter) drives were gradually cheapened out; at first this was fairly benign, done mainly by making the casings somewhat lighter (and omitting the top damping layer from the cover on many – WD even made many otherwise-identical drives with or without the damper, so that volume customers could choose between cheapness or quietness as their priority – such that some were slightly louder than 2‑ or 3‑platter models), and sacrificing seek speed. After those measures were taken as far as they could go, manufacturers cut back on the important parts.
The hot, humid Australian climate seems especially hard on weaker HDDs (c_hegge reported getting only 1–2 years out of later Caviar Blues and their Seagate counterparts, in customer PCs); of course, that didn't stop my ST380011A from surviving here without air conditioning