pcg@cs.aber.ac.uk (Piercarlo Grandi) (12/21/90)
On 14 Dec 90 00:08:30 GMT, lindsay@gandalf.cs.cmu.edu (Donald Lindsay) said:
lindsay> In article <PCG.90Dec12174320@odin.cs.aber.ac.uk>
lindsay> pcg@cs.aber.ac.uk (Piercarlo Grandi) writes:
pcg> There are even drives, magnetic or not, whose mean undetected error
pcg> rate is of the same order as their capacity, so virtually
pcg> guaranteeing that you get an undetected error every time you make a
pcg> copy of them. After all even a fairly respectable undetected error
pcg> rate of 1 in 10^12 is usually expressed in bits.
lindsay> Good point! Creo's 1 TB optical tape holds 10^12 bytes and has
lindsay> "fewer than 1 in 10^12" bit errors. The pessimistic reading is
lindsay> as "fewer than 8 mistakes per reel".
No, the pessimistic say that every time you try to backup a reel you
will get *at least* :-) eight undetected errors (the idea is that since
they are undetected, you don't know). In other words you cannot make a
duplicate of a reel.
Also, how long does it take to duplicate a reel holding 1TB? :-(.
lindsay> Rather than expecting perfection, we should probably expect
lindsay> systems to have selectable, adjustable amounts of protection.
I am starting to think that ECC will become more and more of an
end-to-end issue. If I had a tape unit like above I would make quite
sure that I used some heavvily ECC'ing modification of 'tar' to write
and read to it.
--
Piercarlo Grandi | ARPA: pcg%uk.ac.aber.cs@nsfnet-relay.ac.uk
Dept of CS, UCW Aberystwyth | UUCP: ...!mcsun!ukc!aber-cs!pcg
Penglais, Aberystwyth SY23 3BZ, UK | INET: pcg@cs.aber.ac.ukhank@masscomp.ccur.com (Hank Cohen) (12/26/90)
The discussion of undetedted errors on high capacity storage brings up a question that I have been wondering about for a while. What is the effect of magnetic field "creep" on high capacity magnetic disks and would it be wise to do a zero level dump reformat and restore of such devices periodically like maybe once a year? The question arises because of the well known phenomenon of bit rot on magnetic tapes that leads to the recommendation that long term archive apes be copied once a year. Thanks Hank Cohen