dslreports logo
site
 
    All Forums Hot Topics Gallery
spc

spacer




how-to block ads


Search Topic:
uniqs
493
share rss forum feed


Ugly
Fishy Cool Bird

join:2001-12-12
The Meadow

Is this a credible scenario: MTBF x Eight harddrives. Really?

I read this article, »www.realclearpolitics.com/articl···ils.html and was stuck on a question that can only be answered with mathematics.

Namely, what is the calculation of probability (or chance of actual occurrence) for eight harddrives (L.L's plus seven others, see article) all crashing INDEPENDENTLY by random chance within a short period of time and so close in time to the others?

That is so vanishingly improbable, as to be for practical purposes, almost impossible.

Thus, one asks a thoughtful followup question.
Is there some failure mode that exists, wherein a server malfunction can cause many connected PCs to suffer harddrive crashes as a direct result? ~ Is that a credible scenario? ~ Please explain?

• See in article, "Seven other officials connected to the alleged targeting of political groups also had crashed hard drives, as the agency’s server collapsed."

As I read this, it appears to imply the multiple harddrive crashes were caused by some issue within the server. Is that possible?

Note to mod: If better in another forum, please advise? This forum chosen for the presumed high levels of technical competence and maturity of the typical forum participants.

Note to all: please NO political comments. This is a question only and does not intent to provoke any moral, political, or legal debate. Period. Full stop. Thanks for not trolling my sincere question with politics.

--
Oh, I love the smell of fish. Guts, rotten, it's all good.

dave
Premium,MVM
join:2000-05-04
not in ohio
kudos:8
Reviews:
·Verizon FiOS

2 edits
Server malfunctions don't cause physical drive failures. A disk crash involves a failure of the head to maintain correct flight height; typically that's caused by contaminants or external physical force.

On the other hand, 8 disks are not necessarily 'independent'. If they are the same model, they've got the same design flaws. If they came from the same vendor, they've got the same QA oversights. If they came from the same production batch, they've got the same short-lived production-line screw-up. There is sufficient anecdotal evidence around to say that "bad production runs" do exist.

Or to put it another way, an MTBF is just that: a Mean time between failures. Some common-mode failure could put the entire set of 8 drives elsewhere on the curve.

What's that got to do with any 'server' though? 8 drives on 8 computers doesn't involve a server. Or if there was a server, there wasn't one drive per one client computer.

But the real question is: does anyone believe in a significant data operation that does not do offsite backups?

Edited to add: the web-site commenters are the usual bunch of rightwing nuts. For the record, I don't believe everything is a big plot. However, the "dog ate my hard drive" excuse does not pass the sniff test, and it can be discussed without the need to posit conspiracies starting at the president.


NormanS
I gave her time to steal my mind away
Premium,MVM
join:2001-02-14
San Jose, CA
kudos:12
Reviews:
·SONIC.NET
·Pacific Bell - SBC
said by dave:

There is sufficient anecdotal evidence around to say that "bad production runs" do exist.

Just another anecdote for the readers, but I did a study of chip failures by date code for a product engineer I worked with at Hewlett Packard. I showed that two date codes, representing two distinct weeks (not consecutive weeks) were failing 100% of the final product tests.
--
Norman
~Oh Lord, why have you come
~To Konnyu, with the Lion and the Drum


Ugly
Fishy Cool Bird

join:2001-12-12
The Meadow
reply to Ugly
This is a very civil and high caliber discussion. Thanks!

One question in reply: There were eight employees whose hard drives crashed.
They most certainly did not start their respective careers at IRS all on the very same day, get issued their PCs all from the same lot on the same day, all with hard drives taken from one production lot, of identical model and manufacturer. ~ That is simply not plausible!

Hence, a commonality among eight hard drives based in being identical but for serial number sounds as nearly impossible and so appears as disproven. That's my take on this.
--
Oh, I love the smell of fish. Guts, rotten, it's all good.

dave
Premium,MVM
join:2000-05-04
not in ohio
kudos:8
Reviews:
·Verizon FiOS
I don't know about the IRS, but my employer doesn't wait to buy a computer for a new employee when the employee starts work. They buy in volume and hand them out as needed.

Also, there is the "hardware refresh cycle". Every N years, the old computers get replaced. Allegedly (I don't do IT planning) it's cheaper this way, I suppose from the point of view of maintenance contracts, operational management, etc.

In short, where I work, a large number of us will have got identical computers that all arrived from Dell on the same day.


novaflare
The Dragon Was Here
Premium
join:2002-01-24
Barberton, OH
reply to Ugly
Some is probable. They likely did for example have identical hds in identical pcs or laptops. No large corp buys comps they are going to issue to employees as needed. They buy a number of them at one time. In other words it is possible for 8 employees to be issued computers from the same lot infact they could even have consecutive serial numbers. This assumes only one fact. That these 8 employees started between upgrade cycles for the company (if i read correctly in the quote in this case the irs.

Now what could cause 8 hds on 8 laptops or pcs to crash at one time. A number of things could cause this bad lot. another is something specific to the environment. This would likely be "dirty power" In other words the office outlets etc have allot of power spikes and dips. High humidity or temperatures (not likely in this case) High amount of dust (again not likely). While exceedingly rare i have seen it a couple times. A bad router (if wired connections) can cause other hardware to fail from some sort of power spike along the cat5 cable from the router. This can cause all sorts of hardware to fail including harddrives video cards network cards obviously as well as ram, cpus,motherboards etc. Again this is crazy rare ive seen it maybe 2 or 3 times in 20 years. And also a non issue with wireless. You'd only see file corruption and failed transfers.

My bet is bad batch of hds and or pcs/laptops. As i stated even if they were each hired a few months to a year apart chances are they got the same machines.


Blackbird
Built for Speed
Premium
join:2005-01-14
Fort Wayne, IN
kudos:3
Reviews:
·Frontier Communi..

1 edit
reply to Ugly
Right now, people are witnessing a lot of folks making public statements (I don't mean here) reflecting little "expert" knowledge of computers, servers, networking, and backups. What is missing is exact detail about the time-frame of the actual drive crashes of the 8 people involved, the nature of the "server collapse" (along with its exact time point with respect to the 8 "drive crashes"), and when the eMails were sent. Likewise, there have been statements made that the IRS only keeps backups for 6 months, so a question arises whether that is totally true or to what extent it's true (they seem to be able to pull up my tax records 3 years back). At this point, we (the public) simply don't know these things. But, IMO, at least that much information needs to be expertly determined and made public.
--
The American Republic will endure until the day Congress discovers that it can bribe the public with the public's money. -- A. de Tocqueville

dave
Premium,MVM
join:2000-05-04
not in ohio
kudos:8
Reviews:
·Verizon FiOS
I have heard that some places have a 'data destruction' policy as well as a 'data retention policy'. Since email is notorious for providing evidence in legal discovery, it's probably beneficial to make sure you don't keep old stuff, even if you think you have 'done nothing wrong'. At the very least, if you've got a corporate policy that says everything over N months old is deleted, you can't be accused of selectively losing evidence, since you can point to your pre-existing policy.

So not having email backups is rather different to not having your tax records.

(This might be seen as contradicting my earlier statement asking who believes the lack of backups, but I hadn't heard about the 6-month retention thing then).

Mind you, I am not an expert, even on the internets.


Blackbird
Built for Speed
Premium
join:2005-01-14
Fort Wayne, IN
kudos:3
Reviews:
·Frontier Communi..
said by dave:

I have heard that some places have a 'data destruction' policy as well as a 'data retention policy'. Since email is notorious for providing evidence in legal discovery, it's probably beneficial to make sure you don't keep old stuff, even if you think you have 'done nothing wrong'. At the very least, if you've got a corporate policy that says everything over N months old is deleted, you can't be accused of selectively losing evidence, since you can point to your pre-existing policy.

So not having email backups is rather different to not having your tax records. ...

Indeed, because in that audit I noted, the IRS rep at one regional office stated that they were going to put a particular comment "in the database" for future reference and send an eMail to the other cognizant IRS office which was involved. At the time, that duality struck me as a bit odd, but if the backups have different 'lives', that might explain it.

In any case, with an issue as politically charged as this one, it would serve the public interest if the genuine facts and details about all these things (including the drive-failure incidents) were made officially clear. Otherwise, the pundits on both sides (and their speculations) will make up everyone's minds for them...
--
The American Republic will endure until the day Congress discovers that it can bribe the public with the public's money. -- A. de Tocqueville


Krisnatharok
Caveat Emptor
Premium
join:2009-02-11
Earth Orbit
kudos:12
reply to Ugly
Many federal agencies refresh PCs for the same offices/components at the same time.

I.E. it is very plausible that the IRS's "non-profit investigation office" (or whatever it is called) all received, say, Dell Optiplex 880s at once. It's possible Dell put the same make/model HDD in those computers, and it's possible those HDDs all came from the same production run, etc.

It's also worthwhile noting that at least a few federal agencies have really fucked up email policies. Some only let you keep 1.8 GB on the cloud, but tell employees that they are responsible for keeping emails backed up. This usually means that employees store little on the server, and set up local archives or files that exist only on their C drive. If their computer dies or gets replaced, they lose anything not on the server. They are supposed to burn CDs/DVDs, but few actually do.

And the field techs who replace the equipment don't give a shit concerning the status of the employee's emails, as anything on the C drive is usually not supported by IT anyways. And any PCs that get switched out will have their hard drives wiped and re-imaged or recycled.

Source: first hand knowledge
--
Those who would give up Essential Liberty to purchase a little Temporary Safety, deserve neither Liberty nor Safety.


Ugly
Fishy Cool Bird

join:2001-12-12
The Meadow

1 edit

1 recommendation

reply to Ugly
While it sounds more possible than initially believed (thanks to all posters) for common hard drives to exist among IRS employees, these folks were not, repeat not all together in the same office, and so were NOT subject to a common PC provisioning and support regime.

If that had been true, then a series of near identical hard drives but for serial number would indeed be likely. We agree. However, as these folks were spread about the country in several different offices and states, that appears rather UNlikely.

• Combined together with the different start dates, I still find the common model hard drive failure mode is not well supported by the available evidence.

Of course, such facts are precisely what a robust hearing, via a Special Prosecutor, are intended to find. An interesting topic to be sure.

And bonus; a more scientific, less partisan area for investigation. Nothing is more distateful than accusations of partisan motives displacing sincere concern for the taxpayer & Rule of Law, except for actual occurrences.

When Nixon's administration faced the nearly identical issue of data loss subject to Congressional subpoena forty years ago, the nation was incandescent with rage over the appearance of corruption and cover up. I remember this well. Today, not so much.

Finally, I have direct experience as Krisnatharok See Profile and agree completely. Local IT will happily replace a PC or hard drive and not even ask if valuable data needs backed up first. And US govt. security policy (for secrecy, NOT corruption) requires any recycled memory (of any type or kind) to be wiped many times before disposal.
--
Oh, I love the smell of fish. Guts, rotten, it's all good.


aurgathor

join:2002-12-01
Lynnwood, WA
kudos:1
Reviews:
·Frontier Communi..
said by Ugly:

While it sounds more possible than initially believed (thanks to all posters) for common hard drives to exist among IRS employees, these folks were not, repeat not all together in the same office, and so were NOT subject to a common PC provisioning and support regime.

We probably want to wait until we have some hard data on those hard drive crashes. Did they all die after the investigation was started, or some did before it, for example.

In any case, simultaneous HD crashes are not particularly likely, but not impossible either. A fair comparison would be winning the lottery -- millions try, but only a very few are lucky enough to do it.

For completeness sake, I should mention that it's not particularly difficult to crash a HD.
--
.sig
Expand your moderator at work


FizzyMyNizzy

join:2004-05-29
New York, NY

2 edits
reply to Ugly

Re: Is this a credible scenario: MTBF x Eight harddrives. Really?

my 2 cents

maybe it is sata fail, and they thought it was HD crashed. shrug
»www.anandtech.com/show/4142/inte···s-recall

oh, and my egg roll
death Star hdd
»en.wikipedia.org/wiki/HGST_Desks···failures


novaflare
The Dragon Was Here
Premium
join:2002-01-24
Barberton, OH
LOL i remember those dang things. Me and my mom both had deskstar hds back then. horrid junk... The failure rate on those things was stupid high.