bpulv wrote:
Don't use ANY cloud storage for offsite backup of your photographs! Computer security experts say you should always maintain a minimum of three (3) forms of backup for your photographs. At least one should be off site to protect against loss in the event of fire or natural disaster. The Cloud is not the best choice because your photos will be under the control of a second party that may have technical problems, financial failure or bankruptcy, sale to an unreliable party or may hold your data hostage to rising costs. Furthermore, upload speeds for large RAW files is way too slow and over time causes undue ware to hard drives. My system provides both on and off site redundant backup that is completely under my own control.
My system consists of the following:
1) All storage of photographs is external to the computer’s (iMac) internal drive. The computer's internal drive is only used for speed when actually editing by temporarrly copying files onto its hard drive then copying them back to the backup drives when editing is done and erasing them from the internal drive.
2) Downloads from the camera, editing, final product and primary storage is on a 5Tb Western Digital My Book drive.
a) Each shoot is stored in a folder identified by job name and date.
b) Within each folder are three sub-folders labeled “RAW”, “Edit” and “Final”. All uploads are made to the RAW folder and only copies of the raw files are copied to the Edit folder for editing. The final output that may include JEPG, TIFF, etc. files are transferred to the Final folder.
3) The complete primary storage drive is backed up to a Drobo 5-drive RAID (Redundant Array of Independent Disks). This allows multiple backup copies of the data along with offsite backup. In addition, the Drobo’s firmware automatically corrects any data transfer errors and insures that all backup drives have matched data. The RAID consists of five Western Digital Red Drives, which are designed for this type of service and have an excellent reputation for reliability. Furthermore they are “hot swappable” which means they can be safely inserted and removed from the Drobo without removing power from the system.
a) The Drobo is only powered on when backups are being made. Otherwise, it is powered off. This limits the exposure to possible hacking.
c) Although the Drobo can hold up to five drives, only four are in the unit at any given time. Three of the drives are never removed and provide the basic redundancy.
d) The fourth drive position is used for the offsite backup. Two drives are used for that purpose.Once a week or as necessary, drive 4 is removed from the Drobo and taken to the bank. Drive 5 is removed from my safe deposit box and drive 4 takes it place in the box.
4) Drive 5 is inserted into the Drobo where the data from Drives 1, 2 and 3 are automatically copied to it.
Although this calls for an investment in equipment, it is a one time investment that is economical in the long run.
Don't use ANY cloud storage for offsite backup of ... (
show quote)
With respect, there is some misinformation here concerning cloud storage, and if the description is accurate, a basic misunderstanding of how a RAID system stores data.
First, I beleive I’ve addressed most of the flaws in the anti-cloud argument above, but let me just add that the idea that uplaoading data to the cloud causes undue wear on your hard drive is incorrect. Anytime you copy data from your primary storage to another destination, whether it’s the cloud, an external drive or a RAID, any “wear” is the same - it doesn’t matter what the destination is, it still has to be read from the primary storage. Also, as an aside, most drive failures are unrelated to the head seeks incurred when reading data.
Now to the RAID. You haven’t mentioned what RAID level you’re using, but regardless, the data on the 3 remaining drives cannot be copied in it’s entirety to a single drive which you take off-site (simple math will tell you that’s impossible). You may think that you’re taking all your data “to the bank”, but you’re not - at most, you’re taking only a part of it. You’re essentially “failing” one drive from the redundancy group and then rebuilding the drive when you insert the replacement drive (and in the process, you’re killing the performance of your array while rebuilding the new drive). Let me explain. If you have 4 drives in the array, they can be used in any number of RAID configurations, depending on how it’s configured:
RAID 0 (unlikely). The data is stripped across all the drives for speed - if you lose one drive from the set, you lose all the data since every file is stripped across all the drives.
RAID 1 (mirroring). Two of the drives are used to store data, and the other two are mirrors for redundancy (safe, but you only utilize half the available storage - the rest is used for redundancy)
RAID 1+0 or RAID 10. A combination of 1 and 0. Two of the drives are stripped for speed and the other two are used for mirroring. (Faster and safe, but again only 50% utilization).
RAID 4 (seldom used except by NetApp) 3 data drives plus one drive used strictly for parity data. Each of the data drives contains 1/3 of the data and the 4th drive is used for parity information.
RAID 5 (most likely what is being used). Uses all the drives to store block data and the parity information is rotated across all the drives (the parity information must reside on a different drive from the data). You can lose one drive in the group (what you’re doing when you remove one drive), and the array will continue providing data which it makes up for the remaining drives on-the-fly using the parity information. When you insert your rotated drive from the bank, the array will begin rebuilding the missing data onto the replacement drive. The downside is that if you lose a second drive before the replacement drive is rebuilt, you lose all your data. The other downside is that the array performance drops in half while the array is in “degraded mode” and rebuilding, which can take days with large drives.
RAID 6. Just like RAID 5 except using 2 parity drives so that you can lose a second drive while rebuilding without losing all your data. The downside is more space used for parity that cannot be used for data.
The one thing in common with all these methods is that all of the data never exists on one drive - the one you’re taking to the bank. A RAID array is certainly better data protection than a single drive, but there is no RAID level that will allow you to take all the data from the remaining drives and archive it onto a single removable drive unless you’re only using 1/3 of your remaining 3 drive’s space and mirroring it onto a single drive which would be a very strange and extremely inefficient use of your space. What you should be doing is leaving all the drives in your RAID and keeping a spare, either seperate or online to use in the event of a failure and then backing up or mirroring the RAID to another off-site system or the cloud for DR. By the way, while I understand taking your system down except when backing up, you should know that a majority of drive failures occur when a drive is stopped and then restarted.
If you doubt any of this explanation, take the drive that you remove to the bank, Mount it on different system (exactly what you’d do if your house and the RAID burned down) and see what portion of your files are there (if any). In other words, test your backup, and if you do, you’ll choose a different method.