Backups on Mac OS X

I tend to be quite paranoid about backing up data.   I think many people fail to realise how much “difficult to replace” data may be sitting on their hard drives: digital photographs (not like the old days when one had shoe boxes of prints and/or negatives), financial information, research notes (not just academic but also for hobbies and so on), e-mails (if not stored on IMAP servers — and even if they are I always like to have a back up), and so on.  There’s also a saying that if data is not stored in 3 places, then its not stored at all.

Apple’s Time Machine has been a tremendous boon to make back up easier.  The fact that it is mostly set it and forget it means that once one has bought an external drive that is either permanently connected for a desk top machine or at least regularly connected to a laptop then one is going to be able to recover files with ease (and not just the most recent version, but depending on the space on the Time Machine hard drive, possibly many months or even years).  I use Time Machine with all of my Macs.

The disadvantage with Time Machine is that you cannot boot from a Time Machine disk so that if a hard drive goes bad in your computer you need to reinstall the OS on a new drive and then use the Migration Assistant to recover from the Time Machine backup.  Another approach to backup is to make a clone of a hard drive.  Several applications support this, the best known being SuperDuper and Carbon Copy Cloner.  SuperDuper is free for simple use but the incremental back up features are enabled on payment.  Carbon Copy Cloner is donation ware.   I’ve used SuperDuper for some years now, although I did use Carbon Copy Cloner for a period when CCC had been upgraded for Leopard, but SuperDuper was lagging behind.  The great thing about both of these applications is that they can clone the hard drive so that the clone is bootable.   This is a great fallback if a hard drive fails when you really need the machine.   While running off an external hard drive is slow, it can be a life saver.   I’ve also used this when upgrading the operating system and deciding that I need to revert to the previous version.  In fact I always due a clone before installing software updates just in case there is a problem.  These are also essential tools if one is upgrading one’s system disk.

For my MacBook I have a TIme Machine backup and typically two cloned back-ups — one is on a desktop external hard drive and the other on a portable hard drive.

Of course, if all the hard drives that are used for back up are in the same place, then fire or a break-in may lead to the loss of all of them.  For this reason, one should keep a second copy of a backup elsewhere.  Many people keep a second cloned hard drive at their office (or at a relative’s house) and each week will swap over the drives.

The other option for offsite back up is to use one of the cloud back up services.   The big disadvantage of cloud back up is that it will take forever to do the initial backup given that most of us are on assymetric internet connections so our upload speed may only be 384Kb/s even if we have 8Mb/s download.  Also as some ISPs are imposing bandwidth restrictions and/or charging for bandwidth this can make online back up less practical.   I use Jungle Disk which effectively is a front-end to using Amazon S3 (which I use) and Rackspace Cloud Files.   However, I only back up online a portion of my data — primarily digital photographs and some documents.  Even so with the way digital photos grow, when I return from holiday it can take several weeks to get the photos uploaded because I limit the upload to being hours during the night when my ISP doesn’t count bandwidth usage.  Nevertheless, Jungle Disk works in the background unobstrusively and I’ve found it reliable and easy to use.  There are many similar services such as Carbonite and Mozy.

Another challenge for maintaining data is that when internal drives are not large enough to hold all of our data, then some of it will be moved to external drives.   Also external drives may be more convenient for sharing some sorts of data.   For example, my iTunes library on my iMac is on an external drive, and I shall probably have to split my iPhoto library into separate libraries some of which are on external drives.   In this case, again one needs to ensure that this data is stored in multiple places.  Time Machine can back up external drives but this is mostly going to be impractical unless one has a very large Time Machine drive and only small amounts of data held on external drives.  Another option is to regularly copy or clone the external drives.  Again SuperDuper or Carbon Copy Cloner can be used for this.  Also another possibility, is to use the built in rsync on Mac OS (or perhaps more easily the GUI interface arRsync) to keep two directories in synchronisation with one another.  An alternative method which I have used is to take advantage of the built in software RAID facilities in Mac OS X.   Disk Utility has the capability to set up two external drives (or partitions on external drives) in a RAID configuration.  These can be either mirrored (known as RAID 1, the OS mirrors one partition to the other), or striped (known as RAID 0, the OS splits data between the drives for faster access).  Disk Utility also allows the creation of volumes that span multiple drives (known as Just a Bunch of Disks).  For more information open up Disk Utility and go to the Help entry for Using RAID sets.  I’ve used this on my MacMini for managing media which are mirrored between two hard drives (the internal hard drive is no where near big enough).  I’ve also used it on my iMac with two external drives to mirror my iTunes library into which I progressively ripping my classical CDs (usually at a lossless setting).  This seems to work quite well although I’ve found that it can be tricky making sure that two disks come back online within the appropriate time interval when switching the machine on or rebooting.   If one of the drives doesn’t spring to life within the appropriate time it will be treated as damaged and the OS will continue using the other drive and then if the first one comes up the OS will recopy all of the data again.   (There is a setting in the command line version of Disk Utility to set the timeout — run “man diskutil” in a terminal window.)  This is probably most useful on machines that are never powered down (which is true of my MacMini, and mostly true of my iMac).  However, even in this case there is a need still to have a third copy of the data somewhere, preferably off site.

Probably the most painless way of achieving hardware redundancy but at a cost is the use of a Drobo.   This is a hardware RAID-like solution (“BeyondRAID”) which allow one to use a mixture of disks of various sizes and manufacture and the Drobo manages the drives such that one or more failures (depending on the product) can be tolerated.  It is also possible to hot swap drives.  They are very impressive and I’ve heard many good reports about them but haven’t used one myself, although I think I might well buy one next year as I’m getting to the stage of having so many odd external drives that it becomes difficult to manage.  It is possible to store a Time Machine on a Drobo in which case it provides redundancy for the Time Machine backup — although one still needs to think about an offsite back up of some form.

I mentioned e-mail at the beginning of this blog.   Obviously e-mail can be backed up with the rest of the Time Machine and cloned back ups, but another solution to e-mail is the application MailSteward which can be used to simply to archive e-mail to an external drive, but it also provides some very powerful indexing and database structures to rapidly search those many years of e-mail that all of us accumulate.  Obviously for many individuals retention of e-mails is not necessarily a problem, but anyone running a business will be aware of the need to retain large volumes of e-mail records.




One Response to “Backups on Mac OS X”

  1. davide Says:

    I liked your post. Cheers

Leave a Reply