Fast, Simple and Secure
All we ask of our Photo Storage System
by Rick Stephens
Your storage setup needs to be simple! The old KISS principle, "Keep It Simple Stupid", truly applies here. A complex storage system takes constant monitoring and a lot of luck to be certain it is efficiently making safe backups. Instead of using complex methods, your data files should be in nice big volumes with a simple and automatic backup. The basic requirements for a photo storage system is: Reliability, Performance and Expandability.
Wading through all the possible storage methods and techniques can be a full time job. RAID, JBOD, Backup, Online, Offline, Offsite, Archival... the list of acronyms just keeps on growing. Our goal here will be to cut to the chase, point out some simple solutions that work. It is our belief that the simpler the better, so we will cut out a lot of the geek speak and focus on a solution that you can put in place easily and quickly.
Software Backup versus RAID
RAID1 Mirroring and RAID5
At MacGurus we are not big fans of relying on RAID1 Mirrors for a backup. The big shortcoming of a mirrored RAID is its instantness. With a mirror anything that happens to the main data base drive or drives is copied to the backup. If you corrupt your files on your main drive, you copy the corruption to the backup, instantly. Mirrors are great at providing uptime for a server where dozens or more users rely on availability. A drive failure can't be allowed to take the server offline, so a mirror as one layer of protection makes sense there. That server will have a real backup in place as well as the mirror though.
RAID5 has its place. But it isn't a magic bullet. All that a RAID5 protects from is a drive failure. And it does that with the added risks of a more complex storage system. That complexity adds in additional ways to corrupt or lose your data - particularly corruption of the RAID structure itself. A backup of a RAID5 is just as important as any other storage type. RAID5 needs to be considered just one more layer of protection that can be used in a storage system.
Another requirement with RAID5 is a really good administrative utility to manage the RAID structure and allow a chance of recovery should there be a problem. Most of the cost of a good RAID5 will be in that software - unfortunately this is sadly lacking in many of the lower cost RAID5 solutions. Combined with a backup a good RAID5 storage system can make for a nice all around protected storage system. If you have the money to invest in it, a RAID5 system may be right for you. It is not very commonly used because of the much higher costs of the RAID5 and the additional costs of a backup.
Application based BackUp
MacGurus recommends, and uses for our own photo and general data, an application based backup system. Run on a timely schedule - automatically backing up your data - this is the safest and simplest storage and backup system. A good backup application protects against just about everything, including a drive failure. A backup volume that is only being accessed while the backup application is running, and isolated when not, is safe from many catastrophic events. These protections not only work against the most common 'human error' causes, but also index corruption from a bad installer or an outside corruption introduced via a download.
You must also take into account the weakness of an application based backup. Simply put, if damage occurs to a file, or a deletion, and the backup is allowed to run, the backup will now be updated to match. This can potentially lose you that data. Critical data should be stored in another location to prevent that from occurring. And yeah, that means a third copy of your precious data files should be kept somewhere else. Using tray mounted removable SATA drives or retired Firewire drives, CDs, DVDs, BluRay, Tape, whatever works for you. The only real data protection is multiple copies.
The better backup applications allow thresholds to be set as to how many files can be erased from the backup without an OK from the User. This is a great tripwire that can allow you the oversight before a wholesale deletion occurs on the backup volume during an automated backup. If the backup is processing more files to be erased than are allowed in your preferences then the backup will stop at the erasure and flag you for an approval before proceeding.
Choosing your application
The application we use here is SyncronizeProX. We have been using this one for many years and have recommended it to countless satisfied customers. SyncProX is capable of backing up literally any file, folder, group of files or complete hard drive - to anywhere else. The backup can reside internally, externally or off your computer over the network. We use combinations of both, each computer backing itself up to drive(s) on that computer and then essential data backed up to a backup server as well. The backup server in this case being a retired G4 wth a Burly hanging on it for storage duties.
We use SyncProX to make backups of our data drives to backup drives. We also backup our OS drive on a weekly basis so if something gets corrupted we can be right back up and running with a reboot (perfect when that OS upgrade breaks something important... but nah, that never happens, right?). And we use SyncProX to make daily backup of our RAID volumes to identical size RAID volumes.
The developer of SyncronizProX, Qdea, also has simpler, less costly versions which can only do data files - but not the operating system which takes a lot more specific code to make bootable. There are several other quality applications out there like ChronoSync and DejaVu as well as others, one of the most talked about being SuperDuper which has an added benefit, like SyncronizeProX, of being capable of making bootable backups. There are also applications like Acronis True Image or Genie Backup Manager for our Windows brethren. (I use the Acronis software on my MacPro XP volumes) On Macsm our personal experience is mostly with SyncProX and we have less capability to comment on the use of those other applications. Many of them have a large following of happy users. Pick the one with a user interface you like. Most backup software has downloadable trial versions so you can check 'em out for free.
Simplicity is the key to a good backup. In the picture above is the SyncProX control panel window for backing up a drive named "Local" to a drive named "Backup". Drag and drop simplicity for selecting the volumes to be copied. Set the AutoSync scheduled time, in this case every night at 3:17 AM. Every morning when you sit down to work this window will be up front on your desktop telling you that your backup ran successfully the night before.
We leave the application running all the time. That way when we bring in a camera's flash cards or drives full of pictures from a shoot we can hit the Backup button after downloading and before erasing our cards. Multiple copies are maintained always.
Time Machine versus SyncronizeProX
Time Machine is Apple's backup software shipping with Leopard OS 10.5. It is a wonderful backup application designed to store incremental backups of every hour, giving you the ability to step back in time and look at files before any event where they were erased, manipulated or damaged. The only setup options the user has is to designate the backup storage disk and what data is NOT to be backed up using Time Machine. I want a bit more capability to configure multiple backup schemes. ie: This drive to that drive. And those drives to another set of drives maybe on another computer.
Time Machine is a storage hog. It requires a lot of space to maintain those snapshots of your storage over a long period of time. Without the space, Time Machine starts dropping older files off the protected/recoverable list. Time Machine is not the best thing to backup a large and active photo database. Photo databases are big and somewhat linear for most of us. Taking a snapshot as you labor away creating large layered files is a quick way to have a huge and ungainly backup.
And because Time Machine uses a proprietary storage system, you can't just look at it to determine if the backup is running properly. Checking the veracity of a regular backup volume is pretty easy. Compare the size, make sure the files open properly and are all present and identical on the backup. With Time Machine you must use Time Machine to check the backups. Much more complex and time consuming.
For what it does best, Time Machine is a pretty impressive bit of programming art. However, if the Time Machine database gets too big and complex, which is inevitable on a work computer, then the risks of the TM database getting corrupted seems to become inevitable. I prefer these days to use SuncProX for everything and leave Time Machine to users who use their computer less. If you do use Time Machine I can but recommend you erase it on a regular schedule - maybe every 3 months or 4 months, depending on how dynamic your computer data storage is. This will keep the TM database cleaner and simpler. A key method of preventing a corruption.
The Gurus Certify This Guide Rules