Elizabeth Schwartz wrote: > I'm thinking I'm looking at a specific instance of a generic problem > for which tools probably exist: > > in our efforts to shrink backup expenses until we bleed, I'm needing > to keep track of disk usage metadata and metrics and correlate it with > the ongoing flow of statistics. > > I'm asked questions like: What disk space has been allocated to which > users? What is the rate of growth by type of use? Which disks have > static data (which would show up as low incremental backup volume)? > > At the 20,000 foot view, we have a spreadsheet listing who owns what, > when it was created, when it expires, and other such metadata. > > At the ground view, we have system tools and assorted backup reports. > > We're a very small shop but our storage use is getting to the point > where it's just a wee bit too painful to try to keep track of these > things hand anymore. We can't pay megabucks for enterprise-scale tools > but considering what we spend in backup we could spend a little to > save more. Primary file servers are solaris 10 running zfs; other > servers are solaris, linux and windows; we do NOT control the backup > server but the folks who do are great about running reports *if* I > know what to ask for. > > Any thoughts? Before I start trying to invent this from scratch? > > Solaris: have you tried the nee Sun SE toolkit, now It's available on sunfreeware.com and at setoolkit.org
One of the many things it tracks is disk partition size allocations. Depending on how you do your user volumes (we do 1 per user on zfs because zfs is conducive to that), you could easily track it out of the box with Setoolkit and orcallatorl.se right out of the box. As far as your tape people, you should be able to ask them for reports about daily rate of change for all filesystems for trending purposes, as well as the incremental level that you're using. You'll have to correlate the differences between the incrementals yourself, since, as I recall, you had a relatively convoluted series of incrementals. You should be able to take the difference between your levels, assuming they are sane, and trend them against other backups to get an average among days or over a period of time. Then you could compute your median, variance, and standard deviations to look for unusual spikes. I'm not sure you have to spend money for this sort of trending (if you did, I wouldn't know exactly who to look at) _______________________________________________ Tech mailing list [email protected] http://lopsa.org/cgi-bin/mailman/listinfo/tech This list provided by the League of Professional System Administrators http://lopsa.org/
