does this mean ...
That at the heart of the matter,
their platters have got fatter.
Make room for more chips,
and make bigger ships.
(putting the spin into disk storage)
Western Digital is buying Hitachi Global Storage Technologies for $4.25bn in a friendly takeover – so much for a Hitachi GST IPO. The money comes as $3.5bn cash and 25 million WD shares, worth $750m at a $30.01 share price. It will be funded from WD's cash and about $2.5bn of debt. The combined company will be much bigger than …
I generally feel like these anti-competitive mergers should not be allowed. We need more choices and more freedom, and the competition also drives new and improved products. Too many companies just take the easy way out, give up, and sell out. Why try harder when you can cash in your chips so easily?
However, I think the other side is also broken. Too much success from one company should not be penalized, but the company should be required to reproduce. I think there should be a limit at 40% of the market when you have to split your company into two halves and compete against each other (and of course continue competing against the other companies you've already been beating). That's estimating that the optimal number of choices for freedom is at least five, though I don't know of any concrete research on the topic. Obviously zero or one choice means no freedom, and if you have 100 choices you're too dazzled to find the best choice. (That excessive choice problem is called the 'paradox of choice', by the way.)
...And yet in my own experience, we have had terrible reliability problems using WD drives (MTBF < 2 years), and no problems at all using Deskstar / Travelstar, in a range of desktop and laptop applications, so from my perspective, I hope they don't lower the standards currently held by Hitachi drives, to those of the Western Digital range, as I won't be a happy bunny.
Really? How many WD drives have you had fail that quick? I've been using primarily WD for over 10 years (also used Maxtor, Seagate, Samsung and IBM / Hitachi during that time) and I've only had 1 of the WD drives start to fail (on SMART at least) while others have over 6 years uptime. Maxtor / Samsung just didn't impress with either speed or noise, IBM / Hitachi have the deathstar range and Seagate tended to write data in a manner designed to fragment a drives contents as much as possible. As a result I've stayed with WD, however I fear Hitachi could easily bring their standard of drives down.
It's a sad fact that in a way, *every* hard drive you buy is a prototype. What I mean is that by the time they've been in use in large numbers for long enough to prove their reliability, they'll also be approaching obsolescence and you'll be buying the next generation product (which has not been around long enough to prove its reliability ...)
Accelerated ageing tests can only get you so far. *Every* HD manufacturer has shipped certain drives that were significantly less reliable than they and you hoped. Get over it.
Google published a paper on hard drive reliability - they have enough drives for the statistics to be meaningful, unlike most. They couldn't find any significant difference in reliability between manufacturers. They did see batch problems (drives with similar manufacturing dates and serial numbers being less reliable than they should be) and occasionally problems with a particular model.
And they found that drives were most reliable running at 35C to 40C. Keeping them too cool (below 30C) *reduced* reliability!
A typical RAID array is populated with N drives, all the same model, with near-consecutive serial numbers. This is a bad thing. If one drive fails because of a fault common to the others (design or faulty components) the others are likely to be on the way out, and the likelyhood that one fails during a RAID-reconstruct operation is considerably enhanced.
You do better to use N drives with nothing in common, and the easiest way is to buy one from each of several manufacturers. But we no longer have several manufacturers ... we now have three. At least the drives are getting cheaper. Time to abandon RAID-5 and go mirrored drives (RAID-1 or -10) only?
If you're striping, you really want all the drives to be as close to identical as possible, so you're distribuitng the writes evenly. If one has a faster sustained write, but another has a faster seek time, but a third has the highest data density... I think my RAID controller just preemptively let out the magic smoke.
Perhaps I should have added YMMV ?
My bad experience of IBM/Hitachi was about 6-7 years ago when I got tired of swapping out failed Desk/Travelstars. Granted it may have been a poor run for the company but its hard to look favourably on drives that have given you so many problems. Trust is important and I've never trusted them since.
Biting the hand that feeds IT © 1998–2019