View previous topic :: View next topic |
Author |
Message |
hush Just Arrived
Joined: 04 Nov 2005 Posts: 8 Location: London
|
Posted: Thu Jan 12, 2006 12:54 pm Post subject: norton ghost |
|
|
We use norton ghost to take an image of a server. I checked the size of the used drive space and then the size of the image to see how much it had been compressed. It was 10GB/3GB respectively.
I was surprised that it was compressed so much.
From people’s experience what compression should I expect? Is this level of compression normal?
hush
Last edited by hush on Thu Jan 12, 2006 3:39 pm; edited 1 time in total |
|
Back to top |
|
|
graycat SF Mod
Joined: 29 Apr 2005 Posts: 16777195 Location: London, UK
|
Posted: Thu Jan 12, 2006 1:22 pm Post subject: |
|
|
it depends on what's contained in the image really, as some file types are compressed more than others. Don't forget that "white space" isn't included in the image either.
|
|
Back to top |
|
|
AdamV SF Mod
Joined: 06 Oct 2004 Posts: 24 Location: Leeds, UK
|
Posted: Thu Jan 12, 2006 1:37 pm Post subject: |
|
|
sounds about right to me. note that if you use the command line to create the image you can use the -z switch to specifiy how much compression to use. more compression takes longer to do, but can be worthwhile if you deploy images over a slower network. I would guess for a full server image you would want to be doing this from DVD so this should be less of an issue and the default is probably fine.
|
|
Back to top |
|
|
Secure Lockdown Just Arrived
Joined: 29 Jul 2005 Posts: 2
|
Posted: Thu Jan 12, 2006 2:03 pm Post subject: |
|
|
we don't compress. drive space is dirt cheap these days. the faster imaging speed is more important than disk space savings.
|
|
Back to top |
|
|
AdamV SF Mod
Joined: 06 Oct 2004 Posts: 24 Location: Leeds, UK
|
Posted: Thu Jan 12, 2006 2:47 pm Post subject: |
|
|
1) if compression makes the difference between one optical disk and having to span, compression is worth it. Having to install from a spanned set of DVDs is annoying if you don't need to
2) a well-compressed image will travel across the network faster, which can offset other time costs such as the overhead of decompressing if you are using network based imagiung such as multicasting (or when copying the image over the WAN to another site for someone to be able to use it there)
|
|
Back to top |
|
|
hush Just Arrived
Joined: 04 Nov 2005 Posts: 8 Location: London
|
Posted: Thu Jan 12, 2006 3:48 pm Post subject: |
|
|
AdamV wrote: |
note that if you use the command line to create the image you can use the -z switch to specifiy how |
Is a switch the dos equivalent of an option in unix?
What are the factors that dictate the level of compression?
|
|
Back to top |
|
|
hush Just Arrived
Joined: 04 Nov 2005 Posts: 8 Location: London
|
Posted: Fri Jan 13, 2006 4:07 am Post subject: |
|
|
hush wrote: |
What are the factors that dictate the level of compression? |
I decided to do some research into my own question. What I found was interesting so I have written a bit about it below:
The level of compression or to use the vernacular ‘file reduction ratio’ is dictated by not only the file size but also file type and the compression scheme.
“Size of files after compression” does not dictate the level of compression as the former is a consequence of the latter.
The compression scheme can be split into two sub category’s
i) Lossless
ii) Lossy
A lossless scheme as implied by the name decompresses with no loss to the original data. A lossy on the other hand does.
One of the more common lossless schemes is the Lempel and Ziv algorithm (named after the creators). This works by means of pattern reoccurrence and recognition (be it in code, text or any other format); reoccurring patterns are identified within the files and flagged. This flagging of data requires less space then the data itself thus resulting in a compressed metadata file. During decompression the flags are simply replaced by the corresponding pattern resulting in a file identical to the original.
Lossy is a term not only pertaining to compression but also file formats, an example of this is mp3 or avi. To illustrate- there are sound ranges that humans can not hear, these are removed from a music track (among other alterations) to produce an mp3. The removing of redundant data results in a smaller file i.e compression.
The file size is a factor as patterns can be more effectively identified and utilised in a longer string of data- linking back to the Lempel and Ziv algorithm.
The Lempel and Ziv algorithm is also the reason why file type affects compression size. Picture and music files are more sporadic then text files and code (for example) in terms of data patterns resulting in a poorer ratio.
hush
|
|
Back to top |
|
|
Secure Lockdown Just Arrived
Joined: 29 Jul 2005 Posts: 2
|
|
Back to top |
|
|
AdamV SF Mod
Joined: 06 Oct 2004 Posts: 24 Location: Leeds, UK
|
Posted: Fri Jan 13, 2006 11:35 am Post subject: |
|
|
In terms of ghost or zip or other common programs which compress stuff, the most obvious things which don't compress well are things which are already compressed - most picture, music and movie formats (plus any kind of archive, zip, rar, cab etc). Less than 2% compression is typical, often zero.
(so a server install with an I386 directory of all the cabs in won't compress very well)
Things with fairly primitive compression such as run-length encoding (PCX files, if I remember rightly) will still compress further using more advance algorithms, around 5 to 10%, depending on the file.
Next up is usually code stuff, exes, dlls etc. (again, a server has a relatively high proportion of this stuff)
Most easily compressed is usually document formats, you can often get word docs for example to compress to less than 10% of the original size.
|
|
Back to top |
|
|
hush Just Arrived
Joined: 04 Nov 2005 Posts: 8 Location: London
|
Posted: Fri Jan 13, 2006 1:49 pm Post subject: |
|
|
Thanks for the link Secure Lockdown.
Eliza or should I say AdamV- under 10% compreesion for a word doc, that’s insane! What qualifications do u have? your overall knowledge base seems very impressive.
|
|
Back to top |
|
|
AdamV SF Mod
Joined: 06 Oct 2004 Posts: 24 Location: Leeds, UK
|
Posted: Fri Jan 13, 2006 3:40 pm Post subject: |
|
|
Maybe "often" was the wrong word. In any given bunch of docs, having some at around 10% is not that unusual, particuarly if they have been edited and re-edited a lot. Visio always compresses pretty well too.
No quals in Word or in zipping, but back in '93 when I was writing games software (I was mainly doing ray-traced animations, but some coding) we had three 'fileservers' for swapping files around. Each had 1.44 Mb of disk space, so you got used to using zip quite a lot and gambling how much you could get on one 'server'
|
|
Back to top |
|
|
hush Just Arrived
Joined: 04 Nov 2005 Posts: 8 Location: London
|
Posted: Fri Jan 13, 2006 4:09 pm Post subject: |
|
|
I meant cert’s and qual’s in general. What is your area of expertise?
|
|
Back to top |
|
|
graycat SF Mod
Joined: 29 Apr 2005 Posts: 16777195 Location: London, UK
|
Posted: Fri Jan 13, 2006 4:21 pm Post subject: |
|
|
hush wrote: |
I meant cert’s and qual’s in general. What is your area of expertise? |
Adam / Eliza is what we commonly refer to as a "fountain of knowledge"
|
|
Back to top |
|
|
AdamV SF Mod
Joined: 06 Oct 2004 Posts: 24 Location: Leeds, UK
|
Posted: Fri Jan 13, 2006 4:42 pm Post subject: |
|
|
I consider myself an expert - I used to be a drip under pressure (ex-spurt. Sorry)
Had a wide variety of jobs and environments, pick up stuff as I go along. I know a tiny little bit about a huge number of subjects. Usually just enough to know which bit has gone wrong and which true guru to call on...
|
|
Back to top |
|
|
|