Ghost Solution Suite

Under the question, why the new version of ghost backup restore the slow speed is more than ghost 8.2. From the ghost 11.0 begin to ghost 11.5 is the case, I had to specifically test in a machine used ghost 8.2 and ghost 11.5, a reduction of backup, ghost 11.5 found the time, almost ghost 8.2 to 2 times.

Why ？

I am not quite sure I understand your question 100% correct so let me re-phrase it - are you asking why ghost 11.5 is 2 times slower in restoring an image that ghost 8.2? It is hard to say because we have done our internal tests and 11.5 is slightly fater than 11.0. I dont know whether we have compared 11.5 with 8.2, generally we only compare stats with the previous version, not with all the versions ever created. Having said that I am sure that when 11.0 was released stats were compared against 8.2 and if something like x2 was noticed then we would be fixing it. It might depend on your hardware (slower hard drives and/or memory), whether image is local or restored over the network thus network speed is a great factor and number of other factors. Also ghost 11.0 and later does more post configuration updates than 8.2, checking on the volumes data etc.

Cheers.

I have been using the ghost 8.2, to do the whole disk backup,
Now replaced by ghost 11.5 a lot of really slow,
Please you have time to test,
A new version of the next improvement

I am having the same problem I think?  I have been using Ghost 8.2 to create and restore a local backup of the computer's o/s partition to an image file on to a 2nd partition on the same computer and hard drive by booting down to a virtual partition.  This works very well and the performance obviously depends on the spec of the computer but it is quick.  I have recently tired to update this process so I am using Ghost 11.5 that comes as part of Ghost Console 2.5.  I am please to see that in Ghost Console 2.5 boot wizard Symantec have brought back a feature that PowerQuest Deploy Centre used to have and that is the 'one click virtual partition' feature.  All of my resent tests have been using Symantec's one click virtual partition feature.  When I use Ghost 11.02 or Ghost 11.5 to create/restore this backup it takes about 2 to 3 times longer than if I use Ghost 8.2.  I am using the same switches, computer, hard drives etc and have tried it on lots of different computers with the same results.  Can you shed any light on this?  Thanks in advance.

Does anyone know how to create a bootable USB flash that will allow me to save and restore the ghost images to a folder on the network? I can not make any changes to the server because I work for a school. The server contains many of the student programs used at the school. I am allow to add any files to the shared folder. I need to be able to image gx260 2k, gx270 xp, gx620, and gx755. I want to save money on buying CDs and DVDs.

Thanks,

Rick Pena

 

Hi Rick Pena,

Can you please post your question as a separate post on this forum because it has absolutely no relevance to the topic discussed here.

Hi Gmarks,

One-click VP is similar to Power Quest feature but is much more powerful. Instead of being proprietary it is pure zip archive that you can simply open and add more files as you please and then modify manifest file to reflect your changes. Plus it works on preOS other then DOS.

Speed question. I dont have the answer. As I said we do performance tests but generally from version to version and not from latest version to all previous versions. What filesystem are you cloning and what preOS are you comparing against - NTFS/FAT32/FAT16, DOS/Winpe/Linux etc? Are these old or new machines?

It is possible that some things may be slower (I dont expect 2-3 times), like for instance cloning FAT16 filesystem, which not many people care about. We have improved caching in 11.0 and perhaps this extra caching may affect older machines? You can always try running without cache and compare: -ws -wa etc.

Thanks

Would be interesting to see at least rough results of your speed test.

-split is the switch you want to use to force specific split point. -split=0 tells ghost no splitting. Dont forget that filesystem can dictate what max file size is and so ghost cannot chance that.

This test was using a Dell 745 all in one with an 80GB (7200rpm) HDD and 1GB RAM to ‘restore’ a local image stored on disk 1 partition 2 to disk 1 partition 1 using ‘One-click VP’. The switches used were -clone,mode=pload,src=1:2\image\BACKUP.GHO:1,dst=1:1 -sure –rb. The image that was used was split into 2GB files when created.

I am puzzled about your stats for two reasons:

1. Using cache switches definetely does change pattern in all of our testing and it has no impact on your cloning stats? -ws (no cache on source), -wd (no cache of dest) and -wa (no cache at all).
2. In all our performance tests (depending on preOS, filesystem being cloned, image location etc) ghost 11.5 either outperforms ghost 11.02 by a slight margin (2-4%) or at least performs close enough that it makes no substantial difference. In your run it appears that it is slower by 11.5%!

What filesystems are you cloning and what preOS are you using? Have you tried turning cache off on 8.2 and see whether it makes any difference to your restore speed? Have you tried restoring from a different disk rather than from the same one?

Thanks.

Do you have any other suggestions?  Have you managed to run a simular test?

Thanks

No, I did not. We are busy with development right now. As I said we were comparing 11.0 against 11.5. We dont go back 2-3 releases when we release new version. We have plenty of other testing to perform.
You were not comparing against 11.0, were you?
Underlying disk libraries have changed substantially between 8x and 11x - I do expect to see some differences.

Did you ever come up wth an answer for this?  I am too experiencing much slower speeds pushing or pulling images on 11.5 vs. my older 8.2.  Unfortunately, 8.2 doesn't really work with Windows 7 very well.

I just want a simple, quick, fast DOS Ghost boot CD.

I appreciate you are very busy and did not mean to imply otherwise. As you are busy developing may be you could bear this issue in mind. It would be great if you could somehow improve the speed when creating/restoring an image from one partition to another partition on the same hard disk when using the’ One-click VP’. Thank you.

Out of curiosity (I don't yet use Ghost 11.x so I don't know), are the image files you created made in their native version?  So when you restored using Ghost 8.2, you were restoring a GHO file(s) created using 8.2.  And likewise when restoring using 11.02 or 11.5, the image file(s) you restored were created using those versions?

If Ghost 11.x can read a GHO/GHS file created by Ghost 8.2, see what happens if you restore such a 8.2-created GHO file but use the 11.x client/virtual-partition thing to do it.  I don't think you can go the other way around, as I doubt 8.x can read a 11.x created GHO file, but I suppose you could try that too.

I'm mostly curious to see whether the issue lies with the Ghost file format and whatever data/meta-data might be in that file, vs perhaps it's the 11.x EXE itself that is poorly written or just does lots of "other" stuff as compared to older 8.x EXE's which were "simpler".

Also, just to make sure, the 8.x-created GHO file and the 11.x-created GHO file were both compressed at the same level (or not compressed at all)?  On 8.x it was None/Fast/High, and I don't know what 11.x calls their compression and what's equivalent to what.  I think there is a "-z #" switch you can use to specify exactly what level of compression you want (1-10 I think) and if such a switch exists for 11, and if "level 5" on 8.x is the same as "level 5" on 11.x you might specifify that (or just use None and not worry about it.)

Good luck,
PH

GHO images are not forward compatible but are always backwards compatible - down to v 3.0 I think. There would be no problems restoring 8.2 image creates using 8.2 or 8.0 or 11.0 etc.
Speed issue as I was mentioning might have been due to the underlying disk libraries that have changed when we went from version 8.3 to 11.0.

Although cloning engine has not changed, disk libraries that serve disk access, filesystem access etc. have changed. This may have been the cause although I doubt that it would be much noticable because we always do performance testing from one version to another. We didnt generally do performance testing 2-4 versions back. What I can tell you is that 11.5 that we were comparing against 11.0 was marginally faster in all cases due to new cache. margin wasnt huge 2-10% max but nonetheless it was. I dont know stats between 8.3 and 11.0 let alone 8.2 to 8.3.

Compression switches have not changed and GHO is still using good old zip compatible compression.

If either people still having this problem is still reading this thread, then given what Eugene just posted, I'd say yah.  Create an image using Ghost 8.2.  Then restore that image using 8.2, and this should be "fast".  Then restore that same image using Ghost 11.x, and this should be "slow".  Make sure you boot 8.2 the same as 11.x (ie: don't floppy boot one and virtual-partition boot the other) and the source image Ghost file is in the same location, and ideally the same network drivers (if using the network to access your image) and same machine.  Basically, the only thing that should differ is the ghost.exe and nothing else.   This would isolate the problem down to the binary and the underlying disk/file library routines used in Ghost 11.x.   While I doubt there's anything us end-users would be able to do to fix this, perhaps in the next revision of Ghost, the programmers could look into what changed in the underlying libraries, and see if there's some choke-point that can be removed.  If such a fix is done, maybe even a LiveUpdate rather than a new version would get you back the performance.

While I don't expect testing vs 4+ versions back or anything, I would expect the chain to remain valid.  If 8.3 was comparably  fast as 8.2, and GSS 1.1 was comparble to 8.3, and GSS 2.0 was comparable to GSS 1.1 and GSS 2.5 was comparable to GSS 2.0...etcetcetc.  So *somewhere* there was a break in the performance that should have been noticed.  Admittedly, if the performance was down by only a couple percentage points per revision, then over time after 5 revisions it may be noticeable for those of us who make big jumps in versions.  But from Eugene comments, it appears that there have been no such real performance degradations.

We're still on XP and 2003, so we've had no particular reason to move up in version.  But I know for example when we do someday, it'll probably be because we updated to Vista/7 or 2008 and we wanted a "more compatible/friendly" version of Ghost to go with those OS's.  At that time, chances are, my images themselves will grow significantly and it'll feel like it's taking longer to "do the same thing" but it isn't :)  

PH

You are quite right Paul - it might feel longer but it is not :)

We had to change disk libraries (and this was no small exercise) due to old libraries inability to accomodate new changes required to support new filesystems and other features we wanted to add over time. There is no "botteleneck" to fix - libraries could potentially be slower than the earlier versions although I dont expect it to be that much slower. But it had to be done so that we could add things like virtual disks for instance or better dynamic disks support or better GPT disks support etc.
Also you can now use Winpe or Linux as your preOS and this will clone substantially faster than DOS. You cannot do this with earlier versions.