Gentoo Archives: gentoo-user

From: Alan McKinnon <alan.mckinnon@×××××.com>
To: gentoo-user@l.g.o
Subject: Re: [gentoo-user] Gentoo LiveUSB
Date: Fri, 11 Jan 2008 21:17:16
Message-Id: 200801112311.36891.alan.mckinnon@gmail.com
In Reply to: Re: [gentoo-user] Gentoo LiveUSB by bjlockie@lockie.ca
1 On Friday 11 January 2008, bjlockie@××××××.ca wrote:
2 > > On Friday 11 January 2008, Anthony E. Caudel wrote:
3 > >> 2nd question: I must be dense on this one so someone help me out.
4 > >> Since a USB stick is seen as a hard drive, why can't I do a
5 > >> standard install to it? Is it because until lately they haven't
6 > >> been large enough? I'm thinking of using an 8GB one.
7 > >
8 > > There's a few reasons:
9 > >
10 > > 1. The memory used on those devices has a limited life - about
11 > > 100,000 writes for the good ones and maybe 10,000 for the bad ones.
12 > > With a standard install, frequent writes are the norm (think cache
13 > > and other similar things). This usually ends up at the same spot on
14 > > the disk, meaning your new install will last about a month if you
15 > > are lucky. There are ways around this, for instance how a LiveCD
16 > > does things.
17
18 > Does desktop RAM get constantly refreshed while powered and it
19 > doesn't need to keep any data when not powered?
20 > Is that the difference?
21
22 I'm not sure what you are asking - you're question is poorly framed. So
23 I'll answer what I think you are asking.
24
25 USB sticks use flash RAM and other non-volatile memory technologies.
26 It's not a magnetic disk, it does use transistors but is otherwise
27 completely different to desktop RAM. It's also a whole lot slower.
28
29 The operating system is almost constantly writing stuff to the disk, and
30 not just swap space - many apps cache information and it has to be
31 stored somewhere. This is not a problem for magnetic disks as they
32 don;t really have a limit on the number of times they can be written
33 to. Flash memory does, it stops working after a time. So once you write
34 to a memory cell say 50,000 times, it's probably useless. Trouble is,
35 you have no way of knowing which cells no longer work, so you have a
36 disk with random corruptions. This is usually considered to be a
37 VeryBadThing(tm).
38
39 alan
40
41
42
43 --
44 Alan McKinnon
45 alan dot mckinnon at gmail dot com
46 --
47 gentoo-user@l.g.o mailing list