Data protection - part 2
Why we should speak about data protection and not about backup? This article is not just for service providers but also for data owners.
The traditional idea of "backup" is to copy some data to tape. To be available if necessary.
Customer: We have a Windows fileserver with DDS tape drive and we are running a backup everyday and put the tape on the shelf. We are rotating tapes every 2 weeks. Pretty simple. Why your backup service is so expensive?
The answer is simple. You are not providing simple data storage on DDS tapes laying on the shelf. You are providing to put back exactly specified data in defined time.
Your product is not "data on tape kept for 3 months".
No. Your product is "service allowing that any data from past 3 months will be back in 1 hour and you never lost more than 24 hours of your history".
There is a small difference if car rental company offers you
Car rental by 30$ per day.
Temporary replacement of your car 24 hours daily. The car will be delivered within 1 hour from 1st call in 20 miles distance around your office.
Do you see the difference?
Let's go back to data.
If you are selling "backup" customers expect cheap storage of data an tapes. Are you able to offer option of maximum lost data history 15 minutes? And putting all data back in 30 minutes from 1st call? Yes, it's possible. But not guaranteed. Are you able to backup 100GB in 15 minutes? Probably not. (O.K., possible in some special cases). Can you give 100GB of data back in 30 minutes from tapes? Do you have dedicated tape drive available? Even in case of more parallel restore requests? Is filesystem/database capable to write more than 100MB/s? Really? Seems that some snapshot technology can do it.
What if you have to restore data to DR location? Ups. Seems that some data replication will be required. Will you still call it "backup"?
Yes, technically it's the backup. But slightly more expensive backup than "tape on the shelf".