On 02/20/2014 02:53 AM, Hans Witvliet wrote:
I've been doing some work for an assurance intermediate, so i heard some of the greasy detail people come up with.And these were only the ones being reported to an insurance company.
A) most people don't make back-ups B) those who do, do not do it frequently enough C) they seldom try to restore before disaster strikes D) When needed, they are unavailable (also taken away) E) When available, they are unreadable F) When readable, data is either outdated, or not included.
Do not forget: the sole purpose of doing backups is not to check the item off the list for the auditor. The SOLE purpose of doing a backup is to be able to do a restore. If you can't do a restore, _prove_ you can do a restore, from your backups, then wheat's the point? Oh, it gets worse! Relevant anecdote follows I was doing some network consulting at a firm that does 'surveys', customer satisfaction, product testing and the like. It was a UNIX ship in the days before Linux: SUN, Sequent and other hardware. The new IT manager installed a 'carousel' to do backups. His policy was to take full image backups. Local copies and off site. His policy off 'full image' was to make for fastest possible restore. I thought was was a bad idea, but that was before "Ice age: Dawn of the Dinosaurs": Manny: Sid, whatever you're doing, it's a bad idea. I told him it was a bad idea. I had a bad feeling but couldn't put it into words. As I left he hired a new sysadmin. The guy gave me an uncomfortable feeling. Sure enough, a couple of years later that sysadmin called me asking for a reference. I called the IT manager and asked what had happened. Part of the service the company offered was showing 'improvement', that is comparing this year's results with past years results. But past years were off on backup. One of the analysts asked the sysadmin to retrieve a past result for a specific client, specific year. The sysadmin retrieved the relevant take and untar'd the directory ... only he made a mistake, put a space before the "*". Overwrote the whole machine. Open files continued, that's UNIX. But pretty soon it was apparent something was very wrong. Eventually the previous day's tape was restored, but in effect three days work was lost. The problem was that the backup strategy did not reflect the business processes. It had been optimized for the convenience of the IT department, assuming complete destruction and hence the need for complete restore. I had advised the IT manager not to take full image backups but to take differential or localized backups. The idea of a 'backup only what has changed' is inherent in the way 'rsync' works. In fact 'rsync' would allow full image baseline then incremental. But not to tape. I can't see how 'rsync' can work to tape :-( Which gets me to my next question. Rsync has two areas that confuse me. The first is the '--backup', the second is the '--delete'. The backup seems to be able to keep at least one generation back of the files. Can it keep more than one? How? The '--delete' family is generally confusing. An example: Suppose you have a directory and are accumulating .txt files along with other stuff, but get so many you want to organize them and create TextFiles/ 2012/ Jan-Apr/ May-Aug/ Sep-Dec/ 2013/ as above 2014/ as above and move/sort things. Presumably a 'rsync' will now create that image at the destination. But what about the old top level .txt files at the destination? Presumably without a "--delete-<something>" they will be left there. What other options are there? -- People are more easily led than driven. - David Harold Fink -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org