"Steve Werby" <[EMAIL PROTECTED]> wrote: > the data transfer costs that may be incurred are dwarfed by the time and > cost involved if a disaster occurs and the recovery phase of the > backup/recovery system requires a lot of time to execute manually and the > system is down for an extended period of time.
After sending this I realized that this is pretty vague. What I mean is that if your backup/recovery system requires you to manually recreate each site and user via the GUI (and possibly chown all site files), on a hypothetical server with 100 sites and 400 users it's likely you'll take 10-20 man-hours to manually recreate the sites and users, transfer the files over and set everything up properly. Of course, if you write scripts to store specific data about each site and user and scripts which work in conjunction with the utilities in /usr/sbin/cobalt/ you can automate much of the process, but the point is that without some sort of automation it can be very time intensive. And at least based on my experiencing working with clients of mine who run web hosting businesses on RaQs, a disaster that takes sites offline, especially for a day or more, is likely to result in lost clients. So I tend to think of a working, tested backup/recovery system that minimizes downtime as either insurance or a prerequisite for providing services to customers. It's then a matter of determining how often to transfer backups to an off-server location so that the data transfer costs and age of backups in a worst case scenario are acceptable. This probably seems like common sense to most of us, but most RaQ server admins I do business with either have no backup/recovery system in place, an inadequate system or one that they've never tested. And most feel that nothing bad will ever happen to their server and can't justify the cost of implementing even though in hindsight a disaster resulting in even 2 or 3 lost customers would have justified it. The important thing is to at least have *something* in place to backup files and *transfer them to an off-server location*, even if it's a manual process and is done once in a blue moon. And make sure you know what will be involved to do a recovery and better yet test the process at least once. That's much, much better than nothing. -- Steve Werby President, Befriend Internet Services LLC http://www.befriend.com/ _______________________________________________ cobalt-security mailing list [EMAIL PROTECTED] http://list.cobalt.com/mailman/listinfo/cobalt-security
