Re: OT - 25 hours!

I think that the real point is that some new technologies make other, older technologies obsolete. RAID is what, 20 years old? When it was developed, no one envisioned arrays in which each drive in the array was multi-terrabytes. The technology (RAID) will have to either evolve (Raid level 8?) or be replaced. Exactly when it happens is not the point ... it's coming, however, that point is more or less certain. And it's better to read articles about this and be aware of approaching limitations than to get blindsided with seemingly no warning (when, in fact, warning was available).

Tom's "25 hour" time to build an array is, in and of itself, a good if subtle indication of the fact that limits are being approached.


RnR wrote:
On Sat, 21 Jul 2007 22:05:53 -0400, Barry Watzman
<WatzmanNOSPAM@xxxxxxxxxx> wrote:

I saw an article this week (I wish I could remember where) which basically said that the entire RAID standard was on the brink of total collapse because drives were getting so large the chance of an error (causing failure of the rebuild) while rebuilding a failed drive was approaching 100%.

I wish I could find a link to the article, I think it was in C|Net, ZDNet or Infoworld.

[FOUND IT HERE: http://blogs.zdnet.com/storage/?p=162&tag=nl.e539]

Sounds like something you should read.


Barry, that is interesting reading. I guess I can see both your side
and Tom's side here. For example, lets say you have a 20% chance of
failure which means you have a 80% chance of non failure.

Regardless, I tend to discount this reading LESS than Tom because even
if the probability is less than stated, it still shows where RAID is
vulnerable and if you really want a rock solid system then maybe RAID
isn't the answer or isn't enough. Makes me wonder if you can have 2
RAIDS networked together where one backs up the first one??? Yeah
maybe this is a crazy idea but it just popped into my head.

Thanks Barry. I liked the article.
.

Relevant Pages

  • Bug report for linux-2.6.0-test7: System crashes when I try to start my software raid
    ... the crashes didn't trash my data or raid headers. ... 00:04.1 IDE interface: VIA Technologies, ... 00:0a.0 Unknown mass storage controller: Triones Technologies, ...
    (Linux-Kernel)
  • Re: Building a file server - advice please
    ... > So if you have a 1% failure rate in one year of operation, ... > 4.9% chance of losing at least one drive in one year. ... >> important reason to use a RAID system. ... >> drive failing and still work, the odds of a system failure with 5 drives ...
    (comp.os.linux.setup)
  • Re: OT - 25 hours!
    ... basically said that the entire RAID standard was on the brink of total ... collapse because drives were getting so large the chance of an error ... failure which means you have a 80% chance of non failure. ...
    (alt.sys.pc-clone.dell)
  • RE: OT - 25 hours!
    ... pending failure, I suspect I'll run for years without problem. ... thought about just making it a giant Raid0 array, ... I think that the real point is that some new technologies make other, ... RAID is what, 20 years old? ...
    (alt.sys.pc-clone.dell)
  • Re: can redhat support very big partition?
    ... Partitions cannot be created across physical drives. ... I presume you should be interested in RAID about which I know nothing. ... technologies. ...
    (linux.redhat)