[GLLUG] CD-RW Packet Writing

[Marr]

On Friday 15 October 2004 12:27pm, Matt Graham wrote:
> Marr wrote:
> > On Thursday 14 October 2004 04:26pm, Matt Graham wrote:
> >>> At last week's meeting, someone
> >>> asked about packet writing on optical media.
> >>
> >> "Don't bother" would be my advice. If you have a few M to
> >> transfer, USB keychain drives typically make more sense than CD-Rs.
> >
> > USB keychain drives definitely have their place -- I use them too, for
> > different reasons. But there's obviously no way they can match the CD-RW
> > disc in media costs.
>
> Flash memory is supposed to have at least 100,000 write cycles before it
> dies. CD-RWs have fewer than 1000 write cycles and maybe as few as 10,
> depending on media quality.

That's mostly true but misses my point -- I can give away (or inadvertently 
lose) CD-R[W] media and not feel any financial pain -- not so for USB flash 
RAM drives. The extra longevity of the flash RAM is useless if you need to 
give away the device after 'n < 1000' write cycles.

By the way, I've been burning a variety of CD-RW media since 1997 and have 
never encountered media which dies after anything close to "maybe as few as 
10" cycles. I'd demand a full refund on such shoddy media.

> Not to mention the low-speed vs. high-speed
> CD-RW formats (can't write a high-speed CD-RW in a low-speed drive, can
> write low-speed CD-RW everywhere but only at 4x.)

4x CD-RW is perfectly fine with me, when writing the kinds of things that I 
use packet writing for. Super-highspeed drives are overrated, IMHO. As I'm 
sure you already know, none of those drives achieve the stated maximum speeds 
except briefly, under the best of conditions (outer portion of media, etc).

> >> ISO9660 multisession works just fine and has less overhead/hassle
> >> associated with it.
> >
> > If my math is correct, this is only true for around 10 or fewer sessions
> > per disc.
>
> Yeah, that's correct--but that's what, about 60M/session?  USB keychain
> drives start at 64M these days.

I don't understand your point here. My point was that I'd be using far more 
than 10 sessions on a CD, which quickly negates any benefit (less overhead) 
of using multi-session in lieu of packet writing.

> > The things I use packet writing for would use far more than 10 sessions
> > per disc.
>
> What are you *doing*?

As I said before:

   "... packet writing is a very nice way to write small files 
   on a frequent but sporadic basis."

And I should probably have added this (to help you better see my point of 
view):

   ... to easily-transportable, inexpensive media in the absence 
   of any wired/wireless LAN/Internet connections.

Like I said in my first email, I sometimes need something with all the 
benefits of a floppy diskette but with a much larger capacity.

> Right.  I guess my experience is semi-limited, since I've always carried my
> laptop with me since ~2000.

I hope it's just me reading too much into this comment, but it seems 
needlessly sarcastic to me. :^(

> (12 or 32G hard disk >> 4.4G DVD+-R.)  Most of
> the time, I've had access to the larger hard disks on my home machine via
> cable/DSL.  Things of a few M get transferred over the Net.  Things larger
> than that get copied to my laptop disk.  I use CD-Rs and DVD+Rs for
> archival and backup, not as removable media, since fast networks and scp
> make removable media much less valuable IMO.
>
> > > ?  I just wonder why anyone bothers with packet-writing.
> >
> > I can't speak for anybody else, but packet writing is a very nice way to
> > write small files on a frequent but sporadic basis. Having a 13
> > MB/session multi-session overhead to write one or two 100 KB files just
> > isn't tolerable, to me.
>
> 200K is small enough that it's easily transferable even over a 33.6 modem
> connection.

Your assumption of an easily-accessible Internet connection is a huge one, for 
me (and others) and the places I (they) sometimes use a PC. I know it may be 
hard for some to believe that, but it's true.

> > Furthermore, I can easily erase just one file on a packet-written CD --
> > that's just not possible on a multi-session CD.
>
> This is an advantage, but not a compelling one for me.  When I burn
> something, I want to keep it for a few years.  Transient files get stored
> on disk until they get rm'ed (if they weren't worth keeping) or burned (if
> they were.)

Sometimes, deleting a single file in a set is critically important. Maybe not 
for how you use your storage, but certainly for me and how I (and others) use 
my/their storage.

> > A USB flash RAM drive qualifies in some cases, but not always. It just
> > depends on lots of unspoken issues, like segregation of the files by
> > subject and/or sharability of the media (both of which are better served
> > by multiple CD-RW discs than by a single USB flash RAM drive).
>
> Segregation of files by subject?  Isn't that what directory trees are for?

In some circumstances, yes. In others, no. You assume that any user should 
have equal access to the data on the media -- that's not always true. 
Segregation of files by subject, sensitivity, etc becomes important for some 
people, maybe not for _you_.

> Samba/NFS/atalk + Cat5 = better sharability than multiple CDs in most
> cases. (Wait 10 years, and it'll be "all cases".  Your dog wants
> cable/DSL.)
>
> > Bottom Line: I guess, for me, packet written CD-RW (and DVD+/-RW) media
> > provide all the advantages of a floppy diskette (portability, ubiquity,
> > inexpensive media, single-file-delete capability) but with truly huge
> > relative capacity.
>
> Ubiquity?  You still can't count on a random machine having a CD-RW drive.

You don't need a CD-RW drive to read a packet-written CD-RW disc. I think a 
lot of people wrongly assume this, and I think you knew that already.

To clarify, I'm saying that ubiquity of read-ability (as opposed to 
write-ability) is acceptable to me (for now). And, in fact, CD-RW drives 
_are_ generally available in the machines I need to use. I haven't bought, 
built, or recommended a machine without CD-RW capability in over 8 years.

> CD-ROM, sure, but RW still isn't everywhere, particularly in places where
> the machines all came from one vendor and the beancounters were out in
> force.  Joe Homeuser will have an RW if he bought his x86/Mac in the last
> 2-3 years, but Joe Corporateuser won't.  That may change in another few
> years.  USB 1.1 ports and/or some type of connection to the wide Net are
> more common than CD-RWs right now.

Excepting the USB 1.1. part of your comment, that's not the case for me and 
the machines I often use. And, with the advent of inexpensive external USB/
FireWire <--> IDE bridge enclosure devices, that makes CD-RW drives just that 
much more portable and available nowadays.

> > Ultimately, I might (for now) agree with your "Don't bother" assessment
> > about packet-writing under Linux, but only because it's currently too
> > involved, not because it's not useful.
>
> The patch you talked about in your first message will make it less
> involved, no?

Yes, that is true, which is why I said "I might" not "I do" above. :^)

Bottom Line: For reasons I cannot explain, you seem intent on convincing me 
that packet writing on optical media is useless, merely because you have no 
use for it. Nothing you've said (or could say, given the current state of 
technology) will convince me that packet writing on optical media is not 
useful. We'll just have to agree to disagree on this one, OK? :^) Your needs/
experiences are different than mine.

That said, I do appreciate your comments and insights on the matter.

Regards,

Bill Marr