Need BIG help

[daveyoh]

I need to back up data and retain it for 7 years. If I do a differential M-TH and a full every Friday, and I am using a set1 for week1 and set2 for week2. My scheme would be for example: Jan1=set1/day1, Jan2=set1/day2.....jan8=set2/day1,ect. If a folder containing data were created on my set1/day1 tape and deleted before the set1/day2 backup were run. I believe I could not restore that folder once my tape rotation came around to set1/day2 again. I also believe the folder would not have made it to set1/day5 weekly or set4/day5 monthly because it were deleted before these jobs were run and the set1/day2 has now been reused. Would it be true that in this case the folder would only be recoverable if the deletion was discovered before my set1/day1 tape was over written? I believe to ensure all data is recoverable from any given day throughout a 7 year period, I would need to have 1 tape per day for 7 years without overwriting media.
If I wanted to make a second copy of each tape I would need 2 tapes per day for 7 years (astronomical costs in media).
PLEASE help, I am sleepless trying to figure this out.

 

[bjverzal]

What is stopping you from doing a full every night ?

Bill.

 

[daveyoh]

I could do a full backup every night, but in order to be sure the data is recoverable in the senario I left in the original post, I would need 260 daily tapes a year. I would need twice that to have one offsite and one onsite backup. That leaves me with needing 520 tapes a year. I need to keep data for 7 years. At 7 years I would need 3640 tapes before I could rotate to the first one again.The cost of the tapes alone would be $218,400 and would not include offsite backup storage charges. What I was wondering was is there a better way to prove I can recover any data from any given day within a 7 year period?

 

[bjverzal]

If you have a sandbox, choose a random scenario and just try to recover. I could suggest several hypothetical, but the only TRUE way to prove it is to actually do it. Everything else is purely speculation.

Bill.

 

[dkediger]

Just as an aside - but another part of the "recoverable" thought process:

How is this data to be accessed when it is recovered? Is it just raw data tables that can be read generically through ODBC or some other data connection - or is it uniquely formatted to a particular app?

If it is uniquely formatted for an app, will that data/app combination be viable in say, five years. I've got this very scenario for an older accounting program. The program will not run on anything but a pure DOS machine, of which I currently keep one stashed on a shelf. The likely hood of actually getting this combination to work is very slim and getting smaller each day, but as long as the auditors are happy -

 

[daveyoh]

That is a very good point. The data is raw data which is formated to be read by the program that wrote it. I am going to Ghost image the machine as-is now. I have a duplicate of the hardware and if in 5 years the newest version of the app will not read the data, I will Ghost the image to my generic hardware with the version of the app which wrote the original data. I am just going with the GFS backup routine to keep cost down. I do have another question.....I know I can prevent users from deleting files and also event log users who attempt to delete files, but is there a way to redirect the local workstation's recycle bin to a share on the server? This way if data gets deleted before my backup runs on the workstation it will make it to a backup of the server. I know I can always use a third party "Undelete" program, but it seems like a hassle. I wish MS had borrowed Novell's delete.sav feature.

 

[TechJones]

how much data are we talking about, per night.

 

[Workingman]

This project sounds like you need to install an HSM system so that you can access your data and save on tape cost.

Tivoli and Veritas both sell an HSM system that will help you out.

Note, the cost of the software may seem expensive but in the long run, you will save money.

Good luck!

 

[gwichman]

You are correct in your assessment of how you would not have a backup of your folder in your original message. just as if I created a folder on a particular day, but deleted it before the backup ran, I would not have a backup of that folder. There will always be a window during which something can be created and deleted and thus never backed up. This is an inherent weakness in a standard GFS backup scheme.

One thing you could explore is to use a product like Tivoli Storage Manager which does not use a standard GFS backup scheme. With normal backups it makes much better use of media and can retain all backups over a period while using significantly less media then with a GFS scheme. Granted you'll still not capture the backup of an item that gets created and deleted between each backup that is run, however those files it *does* capture via backup can *always* be available without the cost of doing FULL's, thus your media usage is drastically reduced.