Backing up the data on your computers is extremely important. Despite the fact that the reliability of disks have gone up tremendously in the recent years, they do fail and the effect of losing data can be devastating. In this posting we explain 1) what we do to backup files on our servers, 2) the desktop/laptop service that we provide for faculty and staff, and 3) some pointers about backing up data at home.
We use a system called Tivoli Storage Manager (TSM) from IBM to manage all our backups. It is a very versatile system and helps us manage the backup with very little human intervention. The basic premise of TSM is “incremental backup” - which simply means “backup ONLY the files that have changed between the last backup and now”. Obviously, when you start for the first time, all the files will be backed up and subsequently only the files that changed and any new files created since the last backup will be backed up.
All the relevant data on our servers such as dragon, condor, and mail servers are backed up every night. If you signed up for the desktop backup service, the backup starts sometime after 6 PM every night. The exact time is unpredictable because it is a function of the load on the server. Laptop backups begin sometime after 12 midnight. The desktop or laptop that is scheduled to be backed up needs to be on at the time of backup. The status of backup is shown in the top frame of faculty and employee portfolios and the user can look at more details about the backup by clicking on the “Desktop Backup Status” link in the Tools & Links” bucket.
TSM provides hundreds of options, but there are three that are worth mentioning here:
- Exclusions - we exclude files that have no value being backed up. These are typically operating system files and temporary files which, even if we were able to restore, cannot be used.
- Number of copies of a file - We can configure how many copies of a file to keep in the backup. Sometimes, an earlier version of a file is more useful than the most recent one, so having multiple copies is very useful. For example, if you overwrote one of your documents by mistake yesterday and realized the mistake only today. Last night’s backup would have saved the overwritten document. What you would need is the previous night’s version. We typically keep 2 copies of each file and in some cases the files on servers have three copies.
- Deleted File - Sometimes, one may accidentally delete a file on the disk, realize the mistake a few days later and would want the file back. We configure TSM to keep a deleted file for sixty days, anticipating such situations. The choice of sixty days is indeed arbitrary, but increasing the number of days would consume a lot of hardware resources.
If you are a faculty or staff member and have not signed up for this service, we strongly encourage you to do so. Please contact your desktop support staff to learn more.
Backing up data on home computers
Most of us have one or more computers at home for family use and we very rarely backup the data on these machines. Some of us have attempted to backup what we feel are important files on several CDs or more recently on DVDs. The problem with this approach is, unless one is extremely careful about cataloging what is on these disks and keep them safeand easily accessible , this is not going to be a reliable system.
There are many remote backup services available on the internet. If you feel comfortable with the idea of your data being backed up elsewhere, then you might want to consider this. Our recommendation is that go with a reliable and respected company even if it costs more. It is your personal data that you are backing up and you certainly don’t want it ending up in the wrong hands. As an example, you might want to consider AT&T Online Vault which costs $7.95 a month for the first 2 GB and $2 for each additional GB. Please note that we do not endorse any particular remote backup vendor because we do not use them. You need to do the necessary research before picking a vendor.
If you are not comfortable with the remote backup idea, we suggest that you look at purchasing a large external USB disk drive that can hold backups of all your home computers with additional space for future expansion. You can purchase a 250 GB or 500 GB external disk at the Cardinal Technology Center located in Usdan Center. With a disk like this and a versatile backup software, you should be able to make reliable backups. Here are some pointers:
- Buy a large external USB hard disk (250 GB or 500 GB).
- Purchase a good backup software.
- Many backup software, including the one that comes with Windows XP, tend to store the entire backup as a single file in a format that is not humanly-readable. Browsing and restoring files from such a file is cumbersome. Look for a software that lets you copy the files and directories to the backup disk as is.
- Look for a backup software that lets you save only files that have changed since the last backup. It should also provide you with an option that will not remove files from the backup disk automatically if it was deleted from the source (your computer). You can always manually delete any file(s) from the backup disk to save space.
- Make up a realistic schedule to do backups. If you chose the right software, only the first backup will take a long time. Subsequent ones will complete fairly quickly.
- The easiest method is to connect the backup drive to each machine in the house and backup the information. Create different subdirectories on the backup disk for each of your machines.
- If you are an advanced user, you can leave the backup disk connected to one machine and perform the backup across the network rather than moving the hard disk from machine to machine.
- The natural question is, what happens when the backup disk itself goes bad. Of course, to protect against this, you need a second external disk to which you can copy. But, let us be realistic, let us first get the first disk going… So you understand how this whole thing works, and how easy it is to do. Then you can go get that second disk…
We suggest you look at a program like SyncBack for Windows operating system (a few of us use this software at home and like it). It allows you to configure rules for incremental backup, exclude files from being backed up and is relatively easy to set up and use.
Our Mac expert, Todd Houle, says that there is currently not a good software for the Macs except the usual copying from your computer (either full disks or manually selecting a subset) to the external disk. Todd points out that this is one area where the new Mac OS 10.5, called Leopard, excels. Please read about it here.
Your data is extremely important and backing it up is extremely important. Hopefully the pointers that we have provided here would help you get started.
If you already backup your home computers and want to share how you do it, what software you use and any other useful pointers, please add a comment.
One can also back-up with TSM remotely by using the VPN connection.
One comment that I have about the overnight backup is energy use. This schedule required computers to be left on during the night that means that there are a lot of computers left on, using a lot of energy that are not being used. Perhaps ITS should look into a different schedule to save energy and lower operational costs.
Thank you very much for writing.
Your question is a very valid one and especially in light of the most recent blog posting by President Roth, it is very relevant.
We in ITS have taken this question up a few times and because of the complicated nature of the TSM system, we have not been able to do much. For example, moving the backup time to an earlier time is definitely an option. We can schedule the desktop backups at 9 AM, anticipating that most of the desktop backups will finish by 5 PM. However, when the backup kicks-off, the entire hard disk(s) are scanned to determine if any file has changed. This slows the system so much that if you are working on the machine, it will be very annoying.
The other option is to avoid scheduling them, but rely on the user to initiate the backup manually on a regular basis. We do not think this will work - the backups will be irregular and in many cases, the users will forget to do it.
We have also explored a technology called Wake on LAN - what this simply means is that your computer will receive a signal to turn itself on when we are ready to back it up and then will be turned off at the end of the backup. We looked at this a few years ago and decided not to pursue this on various counts - not all computers had this capability, the technology itself was not mature enough and reliable, and most importantly the TSM client and server actually negotiate several times before the actual backup is made. This means, in a span of 6 hours or so, the computer may have to be woken up and shutdown 2 or 3 times and the points of failures therefore increase.
In a recent inquiry to find out what the other colleges and universities are doing, I received an interesting study by Gustavus Adolphus College - https://gustavus.edu/gts/Power_usage_of_various_devices. Basically they measured the energy consumption by computers. Briefly, what they found can be summarized:
* Computing devices use energy even when they are turned off.
* The differential energy consumption between a computer that is in power save mode and completely turned off is minimal.
* A Mac G4 tower used about 120W while on and idling while a dual core P4 only used 85W.
In addition, newer hardware and operating systems are beginning to pay a lot of attention to energy conservation.
One of our staff member is engaged in studying power consumption by the computers on campus and develop some recommendations. We are also pursuing becoming a beta site for Microsoft’s Green Campus initiative through which we will have immediate access to best practices in lowering energy consumption.
We will certainly re-open this topic in ITS and I will write back on our plans.