51–60 of 259 posts
I like to think the flood was a cover for some government buying a mind boggling quantity of hard drives. All for some massive spying operation. Obviously I have zero proof or even the tiniest bit of evidence. Just my personal conspiracy theory.
I also remember the IBM Deskstar 75GXP, which was so unreliable it earned the nickname "Deathstar".
I worked at a small web hosting company back then, and we used a bunch of IBM Deathstar's. They certainly lived up to their name! I disassembled one just to see what went wrong, and I found the cause (of that drive's failure, at least). Here's a list of terms for reference: http://data-recovery-tips.co.uk/wp-content/uploads/2013/09/h... At some point, the bearing that serves as the pivot that the actuator arm and act…
We'd open the case of a crashed cartridge, extract the platter, and hang it on the wall; the crash would produce a rather spectacular sunburst pattern in the magnetic coating.
1) At one point we were experiencing disk failures often enough that we wrote a cronjob to detect drive failures via smartctl and automatically send an email to our hosting company requesting them replace the drive. This saved engineering time, and more importantly reduced time to drive replacement, because
2) On at least one occasion, we had the third drive in a RAID6 array fail before we had rebuilt from the initial failure, leading to loss of the array. We think the increased load of the rebuild increased the chance of subsequent failures. Needless to say recovering from this destroyed all plans (and sleep) for the weekend it happened.
Back in the mid 90s I was working in IT at a midsized company. We used Compaq desktops (this is the Pentium Pro timeframe). The HD in the system where quantum fireball (IIRC). They failed at an amazing rate all in the first 3 months (30%). We called Compaq at one point and said, look we want replacement drives for all of our remaining systems. After much back and forth (stupid on their part as we had 1000+ machines o…
I'm not familiar with data center tools or terminology. What is a tazer in this context?
Earlier quoted context omitted.
Source for this claim? Sounds entirely like unfounded speculation. (Wikipedia page on the facility in question: https://en.wikipedia.org/wiki/Utah_Data_Center )
Yeah maybe it was just for the construction kickbacks and the datacenter is actually empty.
I just added a photo to the Wikipedia article: https://commons.wikimedia.org/wiki/File:ST3000DM001_as_exter...
In 2012 I was working at the Internet Archive. I remember the 2011 Thailand floods restricted the supply of hard disks and pushed up prices. The best deal we could find was buying these Seagate hard disks as external drives at retail from Costco. We picked up a truck load. Then had somebody with a screw driver remove them all from the enclosures and install them in new servers. The error rate was high, the drives did…
The Thailand story was a cover for the NSA buying a huge number of drives for the new data facility in Utah.
How many people work in Thailand's hard drive manufacturing plants? Hundreds? Thousands? How many people in the logistics channels? How many people are vending food to these employees? How many people in the US Government? Oh, and pulling this off would also require co-ordination with Thai Government. That's hundreds of people right there.
I also remember the IBM Deskstar 75GXP, which was so unreliable it earned the nickname "Deathstar".
Earlier quoted context omitted.
I'm not familiar with data center tools or terminology. What is a tazer in this context?
IIUC the same as in other contexts, the subtext being that in face of their refusal to replace these doomed HDs before a failure has registered, the failures were made to prematurely happen with a gentle stream of electrons.
To the other comments. We where a good customer that experienced an amazing stupidly high failure rate in a part that others also reported the same issue. The cost of deploying these systems and having them fail randomly broke planning and wasted the teams time and impacted our IT cost metrics. The cost of the support ticket, time and replacing and re-imaging a system was not free. We of course tried very hard to be preemptive in sorting this once we ID’ the issue. The vendor did not care, so another path was taken in order to help express our unhappiness to the vendor which resolve the issue. We paid additional money by paying for a brand name vs for white box with the idea that we would prevent this type of failure.