Path: utzoo!mnetor!uunet!lll-winken!lll-lcc!ames!umd5!uvaarpa!mcnc!rti!bcw
From: bcw@rti.UUCP (Bruce Wright)
Newsgroups: comp.sys.ibm.pc
Subject: Re: 20 Meg versus 40 Meg
Message-ID: <2012@rti.UUCP>
Date: 5 Feb 88 04:23:43 GMT
References: <950@its63b.ed.ac.uk>
Organization: Research Triangle Institute, RTP, NC
Lines: 54
Keywords: Hard disks
Summary: What do you want the drives for?

In article <950@its63b.ed.ac.uk>, dougie@its63b.ed.ac.uk (Dougie Nisbet) writes:
> We are intending buying a PC clone with 40 Megabytes of disk space.
> Is it more sensible to buy 2 X 20 Meg, or 1 X 40 Meg drives? I have
> heard rumours that the 2X20's run faster than 1X40. Does anyone know if
> this is in fact the case? Any other reasons for/against would also be
> appreciated.
> 
As with many things, there are several sides to this.  The major advantages
of one drive are:

    1)	The large drives tend to have lower average access times than the 
	small drives (that is, the mechanism itself is higher quality)
    2)	A large drive tends to be less subject to disk fragmentation than
	two small drives (therefore files are more contiguous => less head
	motion to read files)
    3)	The larger drives (as a group) are higher quality drives with a
	larger MTBF.

The major advantages of two drives are:

    1)	If one drive goes down you can always use the other drive
    2)	It is probably cheaper to replace one small drive which goes bad
	than to replace one large drive which goes bad

On the PC it will be rare that two small drives will be faster than one
large drive (exception:  if the large drive's average access time is about
equal to the average access time of the small drive AND the operating system/
disk controller you are running supports overlapped seeks THEN MAYBE the two
small drives can outrun the large one ... note that this excludes all the
major PC system software (MS-DOS, Concurrent DOS, most UNIX systems, etc)
and many disk controllers).  Also, if you are using DOS 2.*, the file
allocation routines are sufficiently stupid that the one large drive would
suffer disproportionately (assuming that you allocated with sector sizes >
512 bytes so that you could use the whole drive).  For the vast majority
of cases the one large drive will be faster.

When we put 40 MB on an old XT we put two small drives in because we were
concerned with system availability at the expense of convenience and speed.
(yes, yes, I know, p(one drive out of two going bad) > p(one drive out of
one going bad) but p(two drives out of two going bad) < p(one drive out of
one going bad) even if the MTBF of the two drives is much lower than the
MTBF of the large drive -- remember I'm talking about whether the system is
useable even in a degraded mode and for our purposes a floppy-only system
even for a couple of weeks would not be a useable mode).

Anyway, I'm not sure we would make the same choice now - the Seagate 225 
drives we used have been a lot of trouble.  If we were doing the system over
again we might go for a real high-reliability single drive (Core, Maxtor, or
something similar), with the possibility of buying a cheap small drive for
when the big drive died. This can of course be considerably more expensive
than even the two-drive solution - but I would definitely recommend _against_ 
going with a single large drive and making it a cheap drive - you are really
asking to take a big hit when (not if) the drive goes bad.

						Bruce C. Wright