Jim:

First, please note that your message contains some sequences (e.g., &#8111
and &#8212, presumably intended to be left and right typographical quotes,
respectively -- not supported in text mode) that badly affect readability.
This can be avoided by creating your message in a message composition
window, rather than in a word processor. Alternatively, you could use the
less economical HTML mode for your messages.

For external storage, kibi, mebi, etc. should be unnecessary. Traditionally,
the capacity of the high capacity storage units used in datacenters has been
specified (correctly) in decimal multiples, regardless of the structure of
the internal memory of the computers themselves. Obviously, this is no
guarantee that people will always get it right. However, if someone says his
installation has a capacity of 50 petabytes, I'm just as impressed as I
would be if he said it had 50 pebibytes (although, in the latter case, I
might be impressed by his erudition).

The maximum theoretical capacity of an 80-column punch card is actually 120
bytes (12 hole positions per column). However, not too many computers were
equipped to read cards in binary mode and there was no data entry equipment
(e.g., keypunches) that could punch them in that mode (nor would it have
been practical to do so). In practice, each column contained less
information than today's byte. Rather, it contained only as much information
as a 6-bit character.

The standard for prefixes to which you refer is an IEC standard. There's a
link to it at http://metric1.org/binprfx.htm.

Bill Potts, CMS
Roseville, CA
http://metric1.org [SI Navigator]



>-----Original Message-----
>From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]
>Behalf Of [EMAIL PROTECTED]
>Sent: Saturday, December 18, 2004 09:09
>To: U.S. Metric Association
>Subject: [USMA:31686] prefixes in data storage
>
>
>I'm posting this partly due to the use of prefixes, and partly
>because I'm sure some of you will find some of the info as
>astounding as I do.
>(For the youngsters on the list, those of us who are 45 or older
>started working on computers when a common storage (and input) media
>was punch cards: about 70 x 150 mm, holding a massive 80 bytes of
>data. That's "bytes," not kilobytes or megabytes!)
>>From the 13 December 2004 issue of Forbes magazine, an article
>titled "Too Much Information":
>"A decade ago the biggest data centers in the U.S. had 10 terabytes
>of storage, and there were only five or ten of them. Today we have
>customers with 2 or 3 petabytes, says Gil Press, senior director of
>open software at EMC, the manufacturer of disk storage systems in
>Hopkinton, Massachusetts. Visa, the credit card company, manages
>more than a petabyte, or 1,000 terabytes. EMC says one of its
>biggest customers, a global retailer, expects to buy 3 petabytes of
>capacity this year . . . . Two years ago the same company bought 300
>terabytes.
>“. . . Storage shipments this year will top 22
>exabytes–or 22 million trillion bytes–of hard disk
>space, says market researcher IDC. That is four times the space
>needed to store every word ever spoken by every human being who has
>ever lived, and it’s more than double the amount sold in
>2002.”
>The rest of the article discusses the challenges of actually
>analyzing or processing that much data.
>If I recall, the IEEE put out a recent standard suggesting we use
>prefixes such as “kibi” and “mebi” to refer
>to binary multiples (kibi = 2^10 = 1024). It sure struck me at the
>time as academic – highly unlikely to actually be adopted. At
>the rate storage is increasing, I think we should be worrying more
>about extending prefixes beyond yotta than whether
>“kilo” is being used “accurately” in the
>computer/binary world.
>Jim Elwell
>
>

Reply via email to