[glb]"Hot-imaging" versus "Cold-imaging"[/glb]

[glb]"PowerQuest" versus "Binary Research"[/glb]

http://img.photobucket.com/albums/v235/El_Pescador/XanToddJ.jpg

Sometime ago, I recall seeing a msg by NBree in the Symantec forum describing how the hot imaging is handled in Windows environment and that from a user point of view the snapshot made by the program is quite reliable. However, there seems to be a very rare possibility of the contents of a file getting changed while the imaging in progress.

It looks like if one wants absolute certainity with the image, one should go for cold-imaging and may be able to sleep well with the confidence that we have a good image for restore purpose if a disaster strikes. While the cold-imaging user interface may not be as easy for the occasional end user customer, I feel the learning curve with cold-imaging is after all worth it in the long run. Customers generally are not sensitive about the backup/restore issues until they get burned by a hardware or software disaster. Personally I have been there in the past.

* [smiley=2vrolijk_08.gif]

There's also a higher chance of generating a conflict. For purposes of this discussion, I'll include a link to my thoughts on the subject, which we've debated:

http://ghost.radified.com/norton_ghost_90.htm

A discussion of this subject may be found in this thread.

I'm almost hesitant to weigh in here, because this topic keeps coming up.  But since this thread is in the Ghost 9/10/S&R board, let me give a few of my comments.


Rad wrote on Mar 26^th, 2007 at 9:44am:

There's also a higher chance of generating a conflict.

I would say that use of DOS and Ghost 2003 creates a higher chance of generating a conflict.

4 years (from 2003 to 2007) is several generations and many light years in terms of PCs and operating systems.  There was DOS, Windows 3, 3.1, 95, 95SE, 98, Me, NT, 2000 Pro, XP, and now Vista.  The real issue is keeping an old operating system and old software working on newer designed hardware and software.  Since 2003, we've had usb, usb2, SATA, PCI express, new motherboards, Core 2 Duo, 64bit, and now Vista which has a completely different boot process than XP.

I'm not saying that you *can't* make Ghost 2003 work today and tomorrow, but the challenge is greater (and the risk of failure greater) each year.  For example,  over the last 2 or 3 years, there have been innumerable threads, questions, posts, about how to get an external chipset and external usb2 hard drive to work with DOS, because you have to have you own drivers for DOS.  I can't even enumerate all the combinations that do or don't work or may or may not work.  

If you are a tinkerer (as many on this forum apparently are) then DOS (and cold imaging) is fine.  I liked DOS 10-15 years ago, and considered myself an expert at tinkering with it.  In 1957 I also liked the '57 Chevy but today I would say there is a higher degree of chance of conflict with a '57 Chevy than a new car.  They are still on the road, and tinkerers love them.

Regarding hot imaging, however, I believe that is a separate subject.  Many large corporations have online databases that are literally online 24/7/365.  Yet they somehow make hot backups keeping our financial and other records safe.  Of course, a purist (tinkerer) could say, "shut them down each day for an hour (maybe even a day for big corps), and take a cold backup" but that's not realistic, and not practical.

Hot imaging and hot backup is built into the design.  That is why vendors are not marketing cold imaging (shut your pc down and boot from this other cd or diskette) products.  They are taking advantage of the current operating system and hardware designs to take a reliable-image-backup.  Whether it uses Snapshot Technology or Volume Shadow Services or some other proprietary method, I personally believe it is less risky that attempting to get DOS programs to work in 2007 on 2007-type hardware and software.

We're already at another DOS/Ghost 2003 hurdle with Vista boot.  I'm sure the tinkerers here will find a way to accommodate that.  Make DOS work with Vista.  How about 64 bit processors?  There will probably be a way to accommodate that as well.  How about new storage devices, flashdrives, ReadyBoost and others that are coming?  There will probably be a way to write or create DOS device drives for that too.

I'm not belittling a tinkerer because I consider myself one.  But I am realistic as to the future.  I try and find current solutions to resolve problems for current PC users on today's hardware and today's software.

Getting stuck 4 years ago or 10 years ago can be nostalgic, but we need to get "back to the future".

(just my 2 cents worth)

Ghost4me, your Reply #4 is exceedingly well stated!

If a user can tinker and get Ghost 2003 to work properly on a PC, there is no reason not to use it.  But, that fact does not constitute an argument against using a “hot imaging” solution (e.g., Norton Ghost 10).

There is simply no evidence – logical or empirical – that a “hot imaging” solution is less reliable than the old “cold image” (DOS) variety.  A few years ago, it was quite understandable to question the reliability of “hot imaging,” since it was a relatively new technology.  Today, however, we are well past that milestone.  It’s not even a question that is asked anymore by the IT intelligentsia, except perhaps by a melancholy few who long for the days when MS DOS was dominate.  (Ah, I remember it well…)

:)

Pleonasm, I had a feeling that for once we would just have to agree to agree!   :)

Hey, Ghost4me – let’s not make it a habit!  I enjoy our ‘interactions’ far too much!

:)

John. wrote on Mar 26^th, 2007 at 1:26pm:

"... I'm almost hesitant to weigh in here, because this topic keeps coming up.  But since this thread is in the Ghost 9/10/S&R board, let me give a few of my comments..."

Actually, to truthfully confess the very reason that I placed nearly-identical posts with such headings and marquees "on-both-sides-of-the-aisle" was not to incite argument but instead to "unravel" the confusing nomenclature promulgated by Symantec through the use of the phrase 'Ghost' or 'Norton Ghost'.  When folks rattled off Version 11 or Version 12, my poor head began to spin because I literally did not know which genera of Ghost was being referred to.

In fact, I am of the opinion that these posts may someday merit a sticky!

EP :'(  

El_Pescador wrote on Mar 26^th, 2007 at 8:57pm:

Actually, to truthfully confess the very reason that I placed nearly-identical posts with such headings and marquees "on-both-sides-of-the-aisle" was not to incite argument but instead to "unravel" the confusing nomenclature promulgated by Symantec through the use of the phrase 'Ghost' or 'Norton Ghost'...

OK; thanks for the chart.  Another example of The Law of Unintended Consequences.   :)

One easy to use rule of thumb is that products labeled "Symantec..." are aimed at corporate customers, while products labeled "Norton..." are consumer products.

Just to confuse matters more, I read somewhere (can't find it anymore) that Norton Save & Restore is actually just a combination of Norton Ghost 10 + Veritas BackupExec (which does the file backups).  From my experience using S&R I would say that the file/folder backup part is definitely an addon and doesn't have the look, feel, or design of Ghost 10.

I agree the version numbers though between Ghost and Symantec are beginning to stumble on top of each other, causing more confusion.

John. wrote on Mar 26^th, 2007 at 1:26pm:

I'm almost hesitant to weigh in here, because this topic keeps coming up.  But since this thread is in the Ghost 9/10/S&R board, let me give a few of my comments.

The facts you stated address the issue of evolution of the situation.

For almost all end users who are not in 7/24 uptime mode, cold imaging should be adequate. However some of us would like to keep uptodate as to what is going on in the technical world where hardware and software innovations take place all the time. May be one day in the near future every one will be using  hot imaging. We will have to wait and see.

For those who are comfortable only with cold imaging for now, Symantec Ghost 11 (enterprise) provides a reasonable answer since it addresses all the issues relating to the internal handling of data by Vista.  Only current hurdle for end users like us is the licensing and I have posted a long msg in Symantec Ghost SS forum suggesting Symantec consider selling single licenses so it is cost effective for individual customers. My hope is Symantec will come up with a way for us to acquire single licenses so that we can own a legit copy of the Ghost 11 (enterprise).

*  [smiley=thumbup.gif]

Rama wrote on Mar 25^th, 2007 at 11:20pm:

Sometime ago, I recall seeing a msg by NBree in the Symantec forum describing how the hot imaging is handled in Windows environment and that from a user point of view the snapshot made by the program is quite reliable.

That's true, I said that and it's very true with one caveat - it applies to Windows XP and above, not to Win2k. When Microsoft introduced the VSS system, in Windows XP the way the default snapshot provider worked was to send an undocumented IOCTL called (from memory) IOCTL_VOLSNAP_FLUSH_AND_HOLD_WRITES down into the storage driver chain for a volume.

This primarily made the volsnap.sys filter driver (interposed between the filesystems and the mass-storage drivers) do its thing which was the "hold part", buffering disk writes in memory until the snapshot was released, when it then caught up by releasing the held writes to the mass-storage drivers. But in addition to this was a special magic part in the filesystem drivers for NTFS and FAT32 - which also knew about this IOCTL and when they saw it, before chaining it on did everything necessary to ensure that the storage was perfectly consistent (at least from the filesystem's viewpoint).

In addition to the above, the user-level component that takes care of the kernel-level magic above uses a COM-based system where any program at all can plug into the snapshot system and ensure that things are logically consistent around the kernel-level snapshot operation - it's a little more elaborate than the power-management APIs that let applications do similar kinds of tidyup when Windows is hibernating, but basically the same idea in principle. It's really hard to fault the architecture of VSS from a technical point of view.

With Windows 2000, Microsoft didn't provide a backport of the filesystem drivers with their part of this adaptation; there was some third-party system that purported to provide VSS support for Win2k, and that's what PQ licensed for what became V2i - and one of the terms they got the original makers to agree to was to not license it to Symantec, which is really the primary reason Ghost32 didn't have VSS support. Anyway, I've never had a chance to look at how the third-party Win2k version  of VSS managed the equivalent to what Microsoft did in Windows XP; whether it was as solid or not I simply don't know.

Now, separately to all of the above is the fact hot-versus-cold is simply not an either/or proposition. It's a continuum; the ability to hot-image allows some neat useful things to be done - for instance, in GSS2 it would allow us to treat the GSS2 console machine in way where it appears under management in itself, alongside the other clients; customers do ask us for that, and it's a perfectly sensible thing to want that we can't really provide seamlessly without using VSS. Enlarging the range of scenarios and letting Ghost fit into more elaborate processes is how we'd have approached VSS, and we wouldn't have forced anyone to use it nor thrown away our capability to take images the old way.

nbree

Very nice input--I appreciate your perspective--thank you!

wrote on Mar 28^th, 2007 at 6:24am:

Rama wrote on Mar 25th, 2007 at 11:20pm: Sometime ago, I recall seeing a msg by NBree in the Symantec forum describing how the hot imaging is handled in Windows environment and that from a user point of view the snapshot made by the program is quite reliable. That's true, I said that and it's very true with one caveat - it applies to Windows XP and above, not to Win2k. When Microsoft introduced the VSS system, in Windows XP the way the default snapshot provider worked was to send an undocumented IOCTL called (from memory) IOCTL_VOLSNAP_FLUSH_AND_HOLD_WRITES down into the storage driver chain for a volume....... With Windows 2000, Microsoft didn't provide a backport of the filesystem drivers with their part of this adaptation; there was some third-party system that purported to provide VSS support for Win2k, and that's what PQ licensed for what became V2i - and one of the terms they got the original makers to agree to was to not license it to Symantec, which is really the primary reason Ghost32 didn't have VSS support. Anyway, I've never had a chance to look at how the third-party Win2k version  of VSS managed the equivalent to what Microsoft did in Windows XP; whether it was as solid or not I simply don't know. Now, separately to all of the above is the fact hot-versus-cold is simply not an either/or proposition. It's a continuum; the ability to hot-image allows some neat useful things to be done - for instance, in GSS2 it would allow us to treat the GSS2 console machine in way where it appears under management in itself, alongside the other clients; customers do ask us for that, and it's a perfectly sensible thing to want that we can't really provide seamlessly without using VSS. Enlarging the range of scenarios and letting Ghost fit into more elaborate processes is how we'd have approached VSS, and we wouldn't have forced anyone to use it nor thrown away our capability to take images the old way.

Thanks for clarifying details about how things are handled inside the Windows and the insight is very valuable for us tinkerers and end users.

Since many of us are still using W2K, at least for me, I feel very safe to stay with cold-imaging which is supported in GSS.

*   [smiley=2vrolijk_08.gif]

I noticed in nbree's (Nigel) profile that he's a Ghost developer ("Principal Software Engineer"). Certainly that gives him huge credibility when it comes to discussing the intricacies of the prgm.

http://radified.com/cgi-bin/yabb2/YaBB.pl?action=viewprofile;username=nbree

http://nigel.bree.googlepages.com/

Welcome.

I can't help but wonder how long he's been developing Ghost. Back when I first started using Ghost (v5.1), it was hard (impossible) to find any help for it. I had to learn the hard way.

Rad wrote on Mar 28^th, 2007 at 6:39pm:

I noticed in nbree's (Nigel) profile that he's a Ghost developer ("Principal Software Engineer"). Certainly that gives him huge credibility when it comes to discussing the intricacies of the prgm. http://radified.com/cgi-bin/yabb2/YaBB.pl?action=viewprofile;username=nbree http://nigel.bree.googlepages.com/ Welcome. I can't help but wonder how long he's been developing Ghost. Back when I first started using Ghost (v5.1), it was hard (impossible) to find any help for it. I had to learn the hard way.

I have been following the symantec ghost forum (mostly used by enterprise users) and I find him very helpful in resolving issues /problems users encounter. In some instances he has been able to get hold of the developer colleague who worked on the original ghost engine to resolve problems. His help to ghost users has been very invaluable and as I have been been around some sticky and serious hardware and software problems with various software, I can easily realize the invaluable help he is rendering to the  Ghost users. I wish more of the software developers get active in the various software boards and forums and there is no substitute for developers' participation. Again thanks Nigel, we are all grateful for your time and help.

*  [smiley=dankk2.gif]  

Rad wrote on Mar 28^th, 2007 at 6:39pm:

Back when I first started using Ghost (v5.1), it was hard (impossible) to find any help for it. I had to learn the hard way.

Indeed, it would have been. You did a great many people a fine service by filling that gap with your excellent guide!

The way things worked with the division between Binary (who were totally focused on R&D) and the marketing side (BRI and the resellers who did consulting and support in various national markets) worked pretty well by the standards of the time, but these days the model for small ISVs to do it all using the web is a bit more well-trodden.

I don't take any credit for the cloning engine, by the way - that belongs entirely to Murray and the other notables at Sprite along with Nigel Pattinson and Andrew Haslam. I was brought in around 1997 by one of the serial entrepreneurs (Gray Treadwell, for whom I'd been a tame geek since the early 80's) who invested in Murray's business and helped grow the business up through to the sale to Symantec.

Anyhow, it turns out I have to correct myself; although IOCTL_VOLSNAP_FLUSH_AND_HOLD_WRITES was still undocumented as of late last year, it appears that it's recently become public, so now you can see the official word on the internal details for yourself. As with most of the design of NTFS itself, it's really just an implementation of concepts that have been pretty well understood for a long time in the database world  - indeed, the great and probably late Jim Gray who was at Microsoft Research for the last few years originally invented many of the key ideas and co-wrote the definitive book on the subject. That doesn't take anything away from the engineers at Microsoft who actually put VSS together, they did a good job with the internals of the implementation.

Nbree, I am confused.  The documentation for Norton Ghost 10 states "Norton Ghost is not integrated with Microsoft's Volume Shadow Copy service" (C:\Program Files\Norton Ghost\Shared\Readme.txt).

Information on the snapshot driver used by Norton Ghost 10 (developed by StorageCraft) is contained in this thread (see especially Reply #9).

Quote:

The documentation for Norton Ghost 10 states "Norton Ghost is not integrated with Microsoft's Volume Shadow Copy service" (C:\Program Files\Norton Ghost\Shared\Readme.txt).

That could mean a number of things, and since I don't have access to their source I can't check, but I'd wager that it just means they use the volsnap.sys ioctl directly, rather than invoking the VSS system through the normal user-mode libraries (it's got an awfully overcomplicated set of COM APIs, and the version of VSS in Win2003 uses a different, incompatible API). Bypassing that simplifies things immensely, and given that very little of anything plugs into the VSS notification system (that's more useful for server code) it's reasonable to do that for WinXP.

wrote on Mar 29^th, 2007 at 6:13am:

You did a great many people a fine service by filling that gap with your excellent guide!

It started with an email to a friend, which contained little more than:

Local > Partition > To Image

.. and grew from there.

Then I noticed I was typing the same email over & over, so I copied-n-pasted it into a web page, and posted it on a friend's site (pre-Radified days).

Once I got this site, I posted it here, and comments started coming in from all over the world, making suggestions.

The rest is history, and the guide continued to grow.

Thanks for your posts, which contain insights not available anywhere else.

Nbree, continuing the conversation from Reply #11, there has been considerable debate among forum members on whether or not "hot imaging" (e.g., Norton Ghost 10) is as reliable as "cold-imaging" (e.g., Norton Ghost 2003).  Your comments imply (and appear to confirm my own perspective) that both are equally reliable for Windows XP - and presumably for Windows Vista, too.  Of course, each image backup approach has some benefits and features that the other lacks - but specifically with respect to reliability, is there any reason to believe that one approach is any more or less reliable than the other?

Thank you.

Quote:

Of course, each image backup approach has some benefits and features that the other lacks - but specifically with respect to reliability, is there any reason to believe that one approach is any more or less reliable than the other?

In the abstract, no I don't think so, although it takes substantial engineering effort to make it so - especially, the cooperation of the filesystem to take on more transactional semantics - compared to offline work. The database world proved all this stuff works three decades ago.

Really, the only thing that's still missing in modern systems is that application programming doesn't support high concurrency and transactional semantics (well, except for functional languages, but despite also being technology proven decades ago they are still a hard sell).

Just to reiterate that hot-versus cold isn't either-or, there's a huge amount of value in having a tool ecosystem in which almost every piece can work either way, which is what we were pursuing in Ghost.  Imaging isn't just about "backup" or "deployment", as much as people like to put it into those categories - it's always a piece of a larger process, and it's nice to be able to give people the choices they need to have the imaging (and other) steps fit into whatever else they want to do.

Nbree, as a Ghost developer and Principal Software Engineer at Symantec, your comments are especially appreciated on this forum.  Your observation that both “hot” and “cold” imaging approaches are equally reliable confirms what users of the former already expected, and may be a key tidbit of insight that allows others who have been hesitant to employ the technology to do so with enhanced confidence.

I am curious:  When you said, “…every piece can work either way, which is what we were pursuing in Ghost” (Reply #21), do you mean that Ghost Solution Suite will be adding (or transitioning to) a “hot imaging” approach?

Pleonasm wrote on Apr 6^th, 2007 at 1:42pm:

When you said, “…every piece can work either way, which is what we were pursuing in Ghost” (Reply #21), do you mean that Ghost Solution Suite will be adding (or transitioning to) a “hot imaging” approach?

I was speaking of our original plans, they got shelved by the PQ "acquisition". It remains technical good sense to add this, customers ask for it, and it's not exactly difficult to do since it's a standard API; however, nowadays it's entirely up to our masters in Orem whether this would ever happen in (real) Ghost. Your guess is as good as mine, basically.

mr. nbree, being a developer of Ghost (one of the coolest programs in the history of software) .. makes you something of a legend. do you feel that way?

if your boot/system drive died .. and if your storage drive contained two back-up images .. one with the file extention *.gho (created by Ghost 2003) and another ending with *.v2i (or whatever file extention Ghost 9/10 produces) .. which would you consider more reliable? .. if you had to bet the farm on the restore working??

Rad wrote on Apr 8^th, 2007 at 6:08pm:

mr. nbree, being a developer of Ghost (one of the coolest programs in the history of software) .. makes you something of a legend. do you feel that way?

It's worth repeating: I don't claim any special credit for the cloning engine. "Success has many fathers" in this case is literally true, along with a reasonable dose of good timing, good fortune, and a healthy amount of plain engineering grind. It's also the case that while you're in the middle of it, the negative feedback you get about your work is rather more direct, pointed, and available (e.g. customer complaints) than the positive feedback, which tends to be delayed and indirect (e.g. sales figures). Plus, there are plenty of obstacles that have to be overcome on the road which aren't very visible from outside.

Those aren't answers, of course. My honest answer is: no, not for a moment. A certain grim satisfaction is about the highest I can muster. Of course, that's not a very illuminating answer unless we were to dig into the why. I'd be happy to discuss that, since what drives us, our values and the yardsticks we use to measure ourselves as human beings are weighty and important topics (ones that you speak to directly in the Rad blog, when you talk about your son). We should perhaps do that outside this particular forum, though; I doubt the Norton Ghost 9 users want to wade through such philosophising.


Quote:

if your boot/system drive died .. and if your storage drive contained two back-up images .. one with the file extention *.gho (created by Ghost 2003) and another ending with *.v2i (or whatever file extention Ghost 9/10 produces) .. which would you consider more reliable? .. if you had to bet the farm on the restore working??

Given that I have the ability and access to fix any problem that could arise in Ghost it's not a question that I can answer in the abstract; I use Ghost, not just because I can but because it's right and proper for us to eat our own dogfood. Indeed, last time I did a boot drive migration at home (earlier last year) it was with the pre-GSS2 development code and helped find a bug.

The same is undoubtedly true of the PQ guys, and I have no doubt that they work every bit as diligently as we do. PQ was justly famous for Partition Magic, and deserved the reputation that brought them. Even as competitors we held their engineering skills in high regard; we and they took and still take different approaches to product design, which is a completely different matter entirely. I happen to feel similarly about most competing products, such as Acronis; I wouldn't criticise the engineering ability of their staff either.

Where all these products differ most is in other qualitative factors; some do quite different things, and so suit different purposes. For instance, PQ's totally sector-oriented approach accelerates incremental backups, while prior to the PQ acquisition it was clear we could extend AutoInstall to use VSS and the NTFS change-tracking facility in Win2k to create similar incremental file-oriented backups (which tend to contain more data and are thus slower to produce, but which would be applied with Ghost on top of any base image).

There's room in the world for all these, and no shortage of reasons to prefer one or another that aren't about technology. Really, most technological arguments about products are emphasising the wrong things: qualitative factors such as "flexibility" and "attractiveness" are harder to evaluate but end up mattering more where it matters most, in how users feel about their experience with a product. We're emotional beings, after all - which ties us back to your question about how I felt about being a developer.

Rad, the question you asked in Reply #24 – whether a .GHO image is more reliable than a .V2I recovery point – was answered in the negative by Nbree in Reply #21, correct?

As a practical matter, to the best of my knowledge, there have been no reported instances whatsoever on this forum or elsewhere of a PC failing to operate properly after a restore from a .V2I recovery point because the state of the machine as captured in the image was “inconsistent” (i.e., the transactional file system activity was improperly managed during the backup process).  That is the only potential point of weakness in a “hot image” approach, and as Nbree noted, the “database world proved all this stuff works three decades ago.”

Pleonasm wrote on Apr 10^th, 2007 at 9:50am:

"... (i.e., the transactional file system activity was improperly managed during the backup process)..."

Then the question becomes: Is easier for a "ham-handed" user to precipitate improperly managed transactional file system activity when using either "hot-imaging" or "cold-imaging" Ghost products, or is it instead a "wash"?

EP :'(

El_Pescador, in theory, almost anything is possible.  I could certainly imagine situations where a user is doing “something exceedingly dumb” while a .V2I recovery point (or .GHO image, for that matter) is being created and thereby cause a problem.  However, even under those conditions, one would expect that the backup process would exit with an error message and a subsequent verify operation would also detect the problem.  (As we have cautioned users many times, it is foolish not to run a verify after a backup.)

wrote on Apr 10^th, 2007 at 6:35am:

[quote author=Rad link=1174840381/15#24 date=1176073688]It's also the case that while you're in the middle of it, the negative feedback you get about your work is rather more direct, pointed, and available (e.g. customer complaints) than the positive feedback, which tends to be delayed and indirect (e.g. sales figures). Plus, there are plenty of obstacles that have to be overcome on the road which aren't very visible from outside.

From an individual end user point of view, Ghost is a classic case of users finding it by word of mouth, not the traditional ad channels. I was told about it by another computer enthusiast when I was visiting a local computer store. And I guess it took some time before Ghost built up its following. From a end user point of view, it has been a godsend in having saved countless hours when something messed up the hardware or software. Our gratitude goes to the developers who fine tuned it to be what it is today.

*  [smiley=2vrolijk_08.gif]

Just to illustrate the point that a .GHO image has the potential to be created in a corrupt state . . .


Quote:

I am trying to create an image of a 12GB Dell 2950 server partition using a Ghost 2003 boot disk.  Every image I create shows as corrupt.  I am able to create images of other servers that are fine.

Source:  Every Image Created is Corrupt

El_Pescador wrote on Apr 10^th, 2007 at 10:26am:

Then the question becomes: Is easier for a "ham-handed" user to precipitate improperly managed transactional file system activity when using either "hot-imaging" or "cold-imaging" Ghost products, or is it instead a "wash"?

The thing is, that the only kind of program which could cause any trouble of any kind would be one that, for instance, when saving a file erases the original before writing a new one. Now, programs that do this are vulnerable to all kinds of problems - if they fail internally, they lose data, if the power goes off, they lose data, etc. - and the total extent of the problem they would cause for a snapshot is exactly the same as in those other cases, which is that one file might end up truncated in that one snapshot.

Now, as it happens there is even a way to deal with almost all these situations; the kinds of programs that do this kind of thing are almost universally single-threaded, which is how Windows stops them causing exactly these problems when, say, suspending a notebook. Because such programs almost always do these operations synchronously, if a backup program were to simply ensure that the currently running programs are dispatching window messages it's pretty much covered.

In fact, this is doable using just one call: BroadcastSystemMessage with the BSF_FLUSHDISK option.

So sure, it's possible to screw things up, but only if you have a program that does long-running, high-risk data-destroying operations on files while still processing window messages. There really aren't a lot of those around though, and people tend not to use such programs on data they value highly.

Pleomasm and Rad

*Hot-imaging*  :-/ *Cold-imaging* reliability--here we go again!


Pleonasm wrote on Apr 4^th, 2007 at 7:10am:

Nbree, continuing the conversation from Reply #11, there has been considerable debate among forum members on whether or not "hot imaging" (e.g., Norton Ghost 10) is as reliable as "cold-imaging" (e.g., Norton Ghost 2003).  Your comments imply (and appear to confirm my own perspective) that both are equally reliable for Windows XP - and presumably for Windows Vista, too.  Of course, each image backup approach has some benefits and features that the other lacks - but specifically with respect to reliability, is there any reason to believe that one approach is any more or less reliable than the other?

Rad wrote on Apr 8^th, 2007 at 6:08pm:

if your boot/system drive died .. and if your storage drive contained two back-up images .. one with the file extention *.gho (created by Ghost 2003) and another ending with *.v2i (or whatever file extention Ghost 9/10 produces) .. which would you consider more reliable? .. if you had to bet the farm on the restore working??

I thought we *agreed* a long time ago that if the software (DOS Ghost 2003 vs Ghost 9/10/Save & Restore) was compatible with one's other software and hardware and did not cause usage errors, then both hot and cold imaging seemed to be equally *reliable*--yes they are different approaches with different functionality--so one person may prefer one approach over the other--but there does not appear to be major *reliability* issues!

I think most of *us* who worry about *hot imaging* are folks who go back some years and used programs in Win 3.x, Win95, Win98, etc. ... years where the promise of being able to multi-task in the Windows environment frequently ended up with a crashed or frozen system--both Windows and the hardware (CPU capacity) were not up to the task of *multi-tasking*!

DOS Ghost 2003 (and it's other DOS cousins) and Ghost 9/10/Save & Restore work for most of the users, most of the time as long as there is not software or hardware conflicts!

Looking over the majority of posts about problems with Ghost in our forum--I would guess that a majority of *reliability* issues is *wetware user error* in not knowing how to use the software correctly (read that--did you read the User Guide?!), or not understanding the terminology of the User Guide (if it was read!)--and not understanding the issues involved in imaging, cloning, and restoring to one's HDD under various scenarios!

So, here you are pestering this poor software developer who has his own corporate work issues to deal with in an attempt to *bolster* your respective biases and have him *take sides*--

Both of you--go to your corners--you need a *time out*, and some *quiet time* to think about this   ;) !



nbree

Ever the diplomatic answer!:


wrote on Apr 10^th, 2007 at 6:35am:

Given that I have the ability and access to fix any problem that could arise in Ghost it's not a question that I can answer in the abstract; I use Ghost (edit by NightOwl--he means DOS Ghost), not just because I can but because it's right and proper for us to eat our own dogfood. Indeed, last time I did a boot drive migration at home (earlier last year) it was with the pre-GSS2 development code and helped find a bug. The same is undoubtedly true of the PQ guys (edit by NightOwl--he means Ghost 9/10/Save & Restore), and I have no doubt that they work every bit as diligently as we do. PQ was justly famous for Partition Magic, and deserved the reputation that brought them. Even as competitors we held their engineering skills in high regard; we and they took and still take different approaches to product design, which is a completely different matter entirely. I happen to feel similarly about most competing products, such as Acronis; I wouldn't criticise the engineering ability of their staff either. There's room in the world for all these, and no shortage of reasons to prefer one or another that aren't about technology. Really, most technological arguments about products are emphasising the wrong things: qualitative factors such as "flexibility" and "attractiveness" are harder to evaluate but end up mattering more where it matters most, in how users feel about their experience with a product. We're emotional beings, after all - which ties us back to your question about how I felt about being a developer.

Thank you, NightOwl, for interjecting a dose of sanity into the discussion.  :) I think you summarized the situation quite well, from my viewpoint:  there really is no rational reason to prefer a “cold imaging” backup solution over one using “hot imaging,” as evaluated in terms of reliability, provided the user has verified in her or his own environment that the selected tool works as expected.

As Nbree stated in Reply #25, we are emotional beings when choosing one backup solution versus another and have “no shortage of reasons to prefer one or another that aren't about technology.”

By the way, for those individuals who may want to learn more about the central role that emotion has in consumer decision making, I recommend the book Body of Truth:  Leveraging What Consumers Can't or Won't Say by Dan Hill.

Quote:

By the way, for those individuals who may want to learn more about the central role that emotion has in consumer decision making, I recommend the book Body of Truth:  Leveraging What Consumers Can't or Won't Say by Dan Hill.

Sounds interesting.

By the way, I wouldn't say that it's irrational to be distrusting of a complex technology like volume snapshotting, particularly when it's a) new, at least to most people and b) underdocumented and hard to understand internally. These are normal reactions to having insufficient information, which is exactly the point where our cognitive machinery needs to turn to our emotions in decision-making.

Part of the problem with the v2i tools is that they were presented as "magic - trust us" (Microsoft haven't done themselves any favours either in regard of VSS, which has some impressively opaque documentation). If, when snapshotting was new, the theory of operation was better explained then it would have been much easier for everyone to become comfortable with it. Giving customers more choice during the transition to using it instead of simply dictating to everyone that the new way was the only way would have helped as well.

NightOwl wrote on Apr 11^th, 2007 at 9:36am:

Pleomasm and Rad *Hot-imaging*  :-/ *Cold-imaging* reliability--here we go again! Looking over the majority of posts about problems with Ghost in our forum--I would guess that a majority of *reliability* issues is *wetware user error* in not knowing how to use the software correctly (read that--did you read the User Guide?!), or not understanding the terminology of the User Guide (if it was read!)--and not understanding the issues involved in imaging, cloning, and restoring to one's HDD under various scenarios!

Most users do not read any of the user guides. I recall an instance of a person who is new to computers spent $3000 and had problems with using it. Called the toll free tech support, and was asked if she read the user manual. She shot back saying that she has spent almost $3000 and after spending all that money should she also read the user guide? We want software much like the video games. Just load and try!

Rama ;D

Nbree, I completely agree:  it is a quite natural human reaction to be initially distrusting of (and have an emotional reaction toward) new technology.  By “not rational” (Reply #33), I simply meant that there is no fact-based or empirical foundation to logically justify the emotion.  That does not make the emotion less “real” or “valid,” of course.

Rama, your point is well taken:  too few people take the time to read the User’s Manual!  When I respond to a technical problem in these forums, I often try to provide a page reference to the manual and point the person to that source rather than answer the question directly.  My motivation is try to encourage more people to read the documentation.

Rad, you are oddly silent here. :-? Do you also agree with NightOwl’s reasonable and thoughtful summary (Reply #32):


Quote:

I thought we *agreed* a long time ago that if the software (DOS Ghost 2003 vs Ghost 9/10/Save & Restore) was compatible with one's other software and hardware and did not cause usage errors, then both hot and cold imaging seemed to be equally *reliable* -- yes they are different approaches with different functionality -- so one person may prefer one approach over the other -- but there does not appear to be major *reliability* issues!

Peace.

Pleonasm


Pleonasm wrote on Apr 4^th, 2007 at 7:10am:

Nbree, continuing the conversation from Reply #11, there has been considerable debate among forum members on whether or not "hot imaging" (e.g., Norton Ghost 10) is as reliable as "cold-imaging" (e.g., Norton Ghost 2003).  Your comments imply (and appear to confirm my own perspective) that both are equally reliable for Windows XP - and presumably for Windows Vista, too.

I believe Ghost 9/10 has been identified as not working under the Vista OS (true?)--so I would have to say those are not *equally reliable* for Vista!

DOS Ghost, by using the *preserve disk signature* switch, i.e. *-fdsp*, based on what I've read (I have no personal experience as yet), allow DOS Ghost to be compatible with Vista however.

Vista and Symantec Ghost 8.x


Quote:

If I perform a DISK backup and then perform a DISK restore AND use the -FDSP switch, I can bring back my Vista 5270 image successfully.

Yes, NightOwl:  you are correct.  Norton Ghost 12 is reported to be compatible with Windows Vista, while Norton Ghost 10 is not.  I should have worded my statement in Reply #20 more carefully.