sarge@metapsy.UUCP (Sarge Gerbode) (10/29/88)
A little while ago, I posted a query on WMRM optical drives. Not having yet received any replies, I thought I'd renew the request. What I'm curious about is -- is there any limit to the number of writes one could reasonably expect to be able to perform on an optifloppy? I would think there would be something analogous to "metal fatigue" involved in repeatedly heating material to its Curie point. Does anyone know if there have been tests showing how many writes at a particular location on the disc one could reasonably expect before the material at that location on the disc breaks down? I would have thought this point would have been thoroughly nailed down by someone. If it's a dumb question, please reply by email and accept my apologies. -- -------------------- Sarge Gerbode -- UUCP: pyramid!thirdi!metapsy!sarge Institute for Research in Metapsychology 950 Guinda St. Palo Alto, CA 94301
ech@poseidon.ATT.COM (Edward C Horvath) (11/10/88)
From article <8646@spl1.UUCP>, by sarge@metapsy.UUCP (Sarge Gerbode): > What I'm curious about is -- is there any limit to the number of > writes one could reasonably expect to be able to perform on an > optifloppy? I would think there would be something analogous to > "metal fatigue" involved in repeatedly heating material to its Curie > point... I posed your question to one of the Sony marketing types at Unix Expo in NYC last week: he said he'd been asked before, and one of the engineers thought about it a bit, filled up a couple of envelope-backs, and estimated the number of writes before breakdown at 1 to 10 million. Probably enough, if the estimate is correct. =Ned=
chavez@sumex-aim.stanford.edu (R. Martin Chavez) (11/13/88)
In article <568@poseidon.ATT.COM> ech@poseidon.ATT.COM (Edward C Horvath) writes: > >I posed your question to one of the Sony marketing types at Unix Expo in >NYC last week: he said he'd been asked before, and one of the engineers >thought about it a bit, filled up a couple of envelope-backs, and >estimated the number of writes before breakdown at 1 to 10 million. > >Probably enough, if the estimate is correct. NeXT uses the Canon drive, not Sony's. R. Martin Chavez Stanford Medical School
pattis@june.cs.washington.edu (Richard Pattis) (11/14/88)
> I posed your question to one of the Sony marketing types at Unix Expo in > NYC last week: he said he'd been asked before, and one of the engineers > thought about it a bit, filled up a couple of envelope-backs, and > estimated the number of writes before breakdown at 1 to 10 million. > > Probably enough, if the estimate is correct. > > =Ned= I sure hope that it breaks down closer to the 10 millionth write than to the first write.
korn@eris.berkeley.edu (Peter "Arrgh" Korn) (11/14/88)
In <568@poseidon.ATT.COM>, ech@poseidon.ATT.COM (Edward C Horvath) said: >From article <8646@spl1.UUCP>, by sarge@metapsy.UUCP (Sarge Gerbode): >> What I'm curious about is -- is there any limit to the number of >> writes one could reasonably expect to be able to perform on an >> optifloppy?... >I posed your question to one of the Sony marketing types at Unix Expo in >NYC last week: he said he'd been asked before, and one of the engineers >thought about it a bit, filled up a couple of envelope-backs, and >estimated the number of writes before breakdown at 1 to 10 million. > >Probably enough, if the estimate is correct. ???? This is considered enough? Lesse here. Assuming I'm using the Optical drive as my main drive (as I might well on the NeXT box), and I'm running my machine constantly (which I would, unless it was in the same room as me at night and made any noise at all), and that I run it with enough jobs to force swapping to disk, we may have a problem here. Let's assume I'm swapping to disk 10 times a second; I'm writing to that same general portion of the disk constantly. If I'm swapping 5 big programs that's twice a second I'm hitting the same portion of disk. 2 * 60 * 60 * 24 = 172800 writes to the same part of disk a day. That would allow me to run the system this hard for anywhere between 6 days and 60 days before I have to replace the disk. That's not a very reassuring. Granted NeXT is using a Canon drive vs. a Sony and I've seen no estimate of this sort on the Canon drive vs. a fourth-hand report on a similar drive made by a different manufacturer. However if anything near these numbers is the case then this generation of optical technology doesn't look ready for Unix to me... Peter -- Peter "Arrgh" Korn korn@ucbvax.Berkeley.EDU {decvax,hplabs,sdcsvax,ulysses,usenix}!ucbvax!korn
steele@unc.cs.unc.edu (Oliver Steele) (11/14/88)
I posted an answer to this in a different thread ("Canon CD", I think),
but it looks like it didn't get read.
In a talk about the 5.25" Sony oppy, by Dr. Robert P. Freese (one of the
patent-holders, and president of Alphatronics, which sells the drives),
he said that his lab had tested the Sony oppies up to 10 million
continuous writes of a single track, without failure (this took about a
month) and that Sony has supposedly tested up to 30 million.
My impression was that this beat magnetic technology, which tends to wear
down and will probably suffer a head crash before that many writes, but
I'd appreciate correction if this is not so.
I'd expect the 3.5" (3"?) Canon oppy to be equivalent.
---------------------------------------------------------------------------
Oliver Steele ...!decnet!mcnc!unc!steele
UNC-CH steele@cs.unc.edu
"It may not be the easy way --
but it's the cowboy way!" - Riders in the Skycaromero@phoenix.Princeton.EDU (C. Antonio Romero) (11/15/88)
In article <7377@pasteur.Berkeley.EDU> korn@eris.berkeley.edu (Peter "Arrgh" Korn) writes: >In <568@poseidon.ATT.COM>, ech@poseidon.ATT.COM (Edward C Horvath) said: >>From article <8646@spl1.UUCP>, by sarge@metapsy.UUCP (Sarge Gerbode): >>> What I'm curious about is -- is there any limit to the number of >>> writes one could reasonably expect to be able to perform on an >>> optifloppy?... >>[Ed got a back of the envelope figure from a Sony type of 1-10M writes...] >???? This is considered enough? >[Pete estimates the heavy swapping would exhaust 10M writes in 6 to sixty days...] >That's not a very reassuring. A couple of things: o I saw an article in either InfoWorld or MacWeek last week (sorry, can't find it now) which claimed that manufacturers had run a drive doing continuous read/write cycles for three months without failure. Not too sure if this was the NeXT's drive, but I think the media are pretty similar... o You're assuming all these writes are done to a fixed portion of the disk. I had the impression that Mach doesn't do its swapping from/to reserved swap space, but rather uses a file. (I'm not sure on this...) This would probably extend life somewhat. o I doubt people would even try to bring something to market if it was as unreliable as this... Think about it. Would it make sense for Jobs, or anyone else, to gamble this way? -Antonio Romero romero@confidence.princeton.edu
mce@tc.fluke.COM (Brian McElhinney) (11/17/88)
korn@eris.berkeley.edu (Peter "Arrgh" Korn) writes: >In <568@poseidon.ATT.COM>, ech@poseidon.ATT.COM (Edward C Horvath) said: >>From article <8646@spl1.UUCP>, by sarge@metapsy.UUCP (Sarge Gerbode): >> >>estimated the number of writes before breakdown at 1 to 10 million. >> >>Probably enough, if the estimate is correct. > >???? This is considered enough? > >Lesse here. Assuming I'm using the Optical drive as my main drive (as I >might well on the NeXT box) [...] That would allow me to run the >system this hard for anywhere between 6 days and 60 days before I have to >replace the disk. A good analysis. Jobs & Co. have said that it is "possible" to run a NeXT box without a hard disk or fileserver, but why anyone would do that? You and I might want to in order to save some money when buying a single machine, but NeXT won't sell to us. In the targeted university market, every NeXT will be on a network and swap over ethernet to a file server. To be sure, they should emphasis this, but hey, pointing out your own bad points has never been the American Way. :-) If a commercial (i.e. single-user) NeXT is ever sold, it will have a hard disk. But that's still several years in the future, if at all... Brian McElhinney mce@tc.fluke.com