We don't back up any user PCs. That is probably putting us at risk of losing some stuff currently, but rather than starting to back them up we're trying to move in the direction of "anything that you can't afford to lose must be stored on the server." I am pretty sure that the cloud solutions that I mentioned below do not do file dedup across systems like BackupPC does with the hard links. However, this becomes less important when you are only backing up a few servers, each with large amounts of data. One of the nice things about the block level dedup is that if you, for example, make a copy of a 15MB spreadsheet and change one cell, it only needs to upload the block containing that changed cell, and all remaining blocks still point to the original copy. Being a healthcare facility we have to follow all of HIPAA guidelines (to the best of our ability.) My main concern that prompted looking at cloud backups is data loss in a total disaster scenario. We can't afford to lose up to a week's worth of data if the building catches fire. I also wanted to make it as automatic as possible. It is inconvienent and puts our data at risk to have someone physically carry something out of the building. Another problem is that all backups are supposed to be encrypted to NIST standards. If we were to carry a hard drive containing backups out of the building, the data has to be inaccessible to anyone that might acquire it. It's quite an unfun problem. From: tbrown@xxxxxxxxxxxxxxxxx To: frgeek-michiana@xxxxxxxxxxxxx Subject: [frgeek-michiana] Re: FW: [News and Security Announcements] TKLBAM, a new kind of smart backup/restore Date: Thu, 9 Sep 2010 13:29:28 -0400 I have casually researched cloud backups and haven’t seen a compelling product so I’m glad to hear about your investigations. Currently I use BackupPC for onsite backups. BackupPC is set to do a full backup of each PC on our network, over the network, once per week and incremental (deltas) backups once per day. Each full backup lasts two weeks before being erased (I can keep more). These are not bare metal disaster recovery backups because of proprietary Microsoft technology which BackupPC can’t or won’t afford to license. Once per week I clone the BackupPC config and data partition to a drive and store the drive offsite. BackupPC relies heavily on rsync and compression to create pools and hard links avoid duplicate data. If there are 42 files which are exactly the same on 42 PCs, BackupPC stores one file and 41 hard links. The hard links are the reason I use Clonezilla to make offsite backups instead of tar, cpio, cp, dd or rsync. Right now I have 42 systems (PCs), 97 full backups and 247 incremental backups stored on about 500GB of a 1TB drive. The PC full backups would total 1671GB before pooling and compression, and the incremental backups would total 367GB before pooling and compression. Servers get the Clonezilla treatment although I could use BackupPC on them as well. Both BackupPC and Clonezilla are open source. Tom -- From: frgeek-michiana-bounce@xxxxxxxxxxxxx [mailto:frgeek-michiana-bounce@xxxxxxxxxxxxx] On Behalf Of Tony Germano Sent: Thursday, September 09, 2010 11:38 To: frgeek-michiana@xxxxxxxxxxxxx Subject: [frgeek-michiana] Re: FW: [News and Security Announcements] TKLBAM, a new kind of smart backup/restore I have been investigating similar solutions to use for offsite backups at work. I'm currently looking at Jungle Disk Business Edition (https://www.jungledisk.com/business/) and CrashPlan PRO (http://b4.crashplan.com/business/index.html) and hope to try them soon. Jungle Disk can use cloud storage at Amazon or Rackspace. CrashPlan backs up primarily to onsite storage for speed and offsite is secondary for disaster recovery. I'm looking at these two because they have a centralized management console for backing up multiple servers. Both of these solutions have block level data deduplication, so if any file has changed you are only uploading a compressed/encrypted block containing the changes rather than the entire file. If the contents of two files are identical, you only back up the unique blocks once and both files point to them. They also support versioning, so you can do point-in-time restores. It looks like TKLBAM uses the rsync algorythm to do delta backups within a file, but I don't see anything saying it does deduplication across multiple files or supports multiple versions of the same file. Still, it's nice to have an opensource player in this market. Tony From: tbrown@xxxxxxxxxxxxxxxxx To: frgeek-michiana@xxxxxxxxxxxxx Subject: [frgeek-michiana] FW: [News and Security Announcements] TKLBAM, a new kind of smart backup/restore Date: Wed, 8 Sep 2010 14:06:43 -0400 If you use Turnkey Linux, this is great news. If you don’t, it is possible (probable?) contributors will adapt the open source TKLBAM tool to other variants of GNU/Linux. TKLBAM is designed specifically for use with TKL servers in the Amazon EC2 cloud, not sure if it works in other scenarios because the authors don’t illustrate other scenarios. Tom -- From: admin@xxxxxxxxxxxxxxxx [mailto:admin@xxxxxxxxxxxxxxxx] Sent: Wednesday, September 08, 2010 08:10 To: tbrown@xxxxxxxxxxxxxxxxx Subject: [News and Security Announcements] TKLBAM, a new kind of smart backup/restore system Dear user, Big news! We've officially unveiled TKLBAM (TurnKey Linux Backup and Migration), a new kind of smart automated backup and restore system that will make it super easy to test your backups "in the cloud", and migrate working systems anywhere in minutes: http://www.turnkeylinux.org/blog/announcing-tklbam Technical documentation: http://www.turnkeylinux.org/docs/tklbam Cheers, Liraz Siri Co-founder of TurnKey Linux Cell: +972-54-2013512 Unsubscribe from this newsletter