You don't work for them by chance?
The more backup solutions available the better IMO.
I am eager to do some testing with VMware Consolidated Backup as well.
Please excuse my pessimism, but this was your first post. Do you work for this vendor or is it just coincidence?
How come NO software company ever mentions the crash-consistent state of doing backups of vmdk files in redo mode? Even VMware doesn't seem to mention it.
Granted, it's no big deal for most servers, but you never hear ANY of these products mention the potential problems with backing up database VMs in such a manner (SQL, Exchange, Active Directory DCs, etc.) It's exactly the same as powering a server off hard, which can lead to serious database corruption (since you lose everything that was in memory at the time of the power-off/hot vmdk backup).
You can get around this by using the built-in application tools to back up the databases to a flat file each night before the vmdk backup. Then, after you restore the vmdk, you can then restore the database from that file.
But no one ever mentions that!
And it's important... How many people actually do full DR tests where they restore all of the vmdk backups they're making?
Nice advert, although it does sound interesting.
I'll give it a whirl once it's released.
I am glad to see you removed your original mocking post of Rob as it was in extremely poor taste. Rob is a great guy and your attack was completely uncalled for, unprofessional and inexcusable.
And no, Rob does not work for me, but I guess you already figured that out. That's probably why you edited your post. Like he stated, we installed our software after he told me he was without backups. A scary position when you have 120+ VMs.
esXpress is my product and Rob was not asked to post here. Not that I was not going to, I was just going to do it on March 15, the official public release date. Otherwise I would have been here first, shamelessly self promoting myself, our product and company.
To rrosenkoetter. In response to your question; "How come NO software company ever mentions the crash-consistent state of doing backups of vmdk files in redo mode?" As a matter of fact we do address this issue as we have performed extensive testing in this area. Of course the software is not even released yet and you have already lumped us into the "But no one ever mentions that!" category. I guess I am not allowed to address this when the software is released. So, in order to make you feel better here is a snip from the manual:
\** A note on Oracle Backups **
While esXpress simultaneously adds REDO logs to all VMDKs associated with a given vitual machine to ensure consistancy across all volumes, we cannot guarantee the kernel scheduler all always honor the request. While our database consistancy tests have been very successful (including our own production environment), our recommendation is to always design the database so that the control files are stored within in the same VMDK file as your Oracle datafiles. This will always ensure datbase consistancy regardless
of delay in REDO application.
And for your question of "How many people actually do full DR tests where they restore all of the vmdk backups they're making?". I don't know. I can tell you that at the Fortune 100 data center that we manage, our entire environment is virtualized. We practice our DR drill every year, restoring EVERY server, not just a select few. Three years in a row with a 100% success rate, VMware is great for DR.
MR-T, thank you for keeping an open mind. I am glad you are willing to give the product a chance instead of throwing stones. I will make sure you recieve a copy on release day.
So, for everyone else, in a week or so, download the software and give it a try. If it does not work as advertised, then bash me & the product.
Ahh, this used to be such a friendly place.
>> I am glad to see you removed your original mocking post of Rob as it was in extremely poor taste. Rob is a great guy and your attack was completely uncalled for, unprofessional and inexcusable.
>> So, for everyone else, in a week or so, download the software and give it a try. If it does not work as advertised, then bash me & the product.
>> Ahh, this used to be such a friendly place.
No need to be so insulted...
I thought it was a plant as well and so would anyone else who's ever spent any time on any message board. First time posters VERY rarely start off mentioning a brand-new product as the bestest greatest thing ever.
I wouldn't call it unprofesional or inexcusable for anyone to question such a post... even in a sarcastic manner...
I would expect such a response, since I'd be willing to bet 90%+ of us thought the same thing (And 99.9% of the time we would be right).
Good luck with your product... If the Deltas work as advertised, it should do very well (do any other "Backup the whole .vmdk file" products offer Deltas?)
Message was edited by:
I apologize for the original post. I realized how mean-spirited it was and didn't mean to have a "knee-jerk" reaction to what I saw as an advert. I had started to see these types of first-time posts and responded poorly. Good luck with your venture and I did take a look at PHD's site.
While esXpress simultaneously adds REDO logs to all VMDKs associated with
a given vitual machine to ensure consistancy across all volumes, we cannot
guarantee the kernel scheduler all always honor the request.
Ken, I feel this sentence lends too much comfort to the casual user who may not understand what that disclaimer at the end means. ESX currently provides no means to simultaneously add REDO logs to multiple .vmdk files. You have to separately invoke vmkfstools or make an API call for each .vmdk file. This means that you have NO control over how much time has elapsed between the application of the REDO logs. On a fairly quiet system, this should not be a problem; however, on a heavily loaded host, you could easily wind up with your .vmdk's out of sync.
In short, there is no way to ensure the consistency of the hot backup of a VM that uses more than one .vmdk file. I know that you are not trying to hide anything, I'd just prefer to see a stronger statement that warns the user of the risk involved...
Best of luck with your new product!
Thats OK I understand, I wasn't insulted. Just had to defend myself.
But I do disagree with you. You can question the authenticity of a post without personaly attacking and degrading the poster, as the now removed original post did.
So to all in this forum that may have been offended that Rob posted here, I personaly appologize. To everyone else, try the software, then beat me up.
As you will see from my posts over the years, besides being consultants, we actually use the software in the data centers we manage. We HAD the same problem as everyone else, to many VMs to little time and space. This software was originally written to allieviate our backup problems, it was not originally a release product by design. It wasn't until we gave a few copies away and saw the response (like Rob's) that we saw the potential.
You have been a great asset to this forum and I personally wish you the best of luck with this. Now if you could put your app it in the form of a small appliance maybe you can win the contest.
Thank you everyone. I did not want to start a flame war, I guess I'm just stressed lately and a little defensive.
ken.cline, I'm going to put up a whitepaper we wrote on maintaining disk and database consistancy across multiple vmdks with redo's.
Steve B, thank you also, and an appliance is in the works, but this has been stressfull enough.
Hopefully it will fill the void for a few users and work for them too.
Well, its dinner time, good night all.
I didn't jump on the new poster...can I get a copy too :O)
We actually just purchased ESX Ranger for our DR solution as its been out for a while now and recommended by quite a few others. I work at a client who doesn't believe in purchasing anything when its been released in its first year.
I def see potential for this product and will be bringing it up at the next team meeting...
What Rob meant was 800GB compressed and backed up. The actual size of his VMDK files are 4.5TB.