| gabriel rosenkoetter on 9 Oct 2003 22:03:02 -0000 |
|
On Thu, Oct 09, 2003 at 02:39:41PM +0000, beldon@speakeasy.net wrote: > Does anyone have any information, experience, or resources related > to using rsync to replicate data volumes on an Oracle database? > Related to this, any (FOSS or not) HA applications? Assuming that you're using raw partitions (or direct I/O files within something like VXFS), this probably won't work. rsync makes no claims about doing the right thing with sparse files; best to assume it does the wrong thing. (Or just test it. > 8. We are a 24x7x365.25 operation-- i.e. we never close. Because > of that, we will have two complete systems in two separate buildings > about one half mile apart, and connected by a gigabit fiber > connection. You can't use rsync to synchronize Oracle DBs without shutting down the DB first. The Oracle DB on the other end will refuse to use the DB files; they'll be out of sync with each other and out of sync with the control file. Unless you've got archive log mode on in your DB, you MUST stop the DB in order to have a useable backup. If you've got archive log mode on and are talking about rsync'ing the dumps out of that... I guess you could, but why not just do it over dblink? It'll probably be faster. EMC-style BCVs would, obviously, also be good for this. I think Hitachi can also do this in hardware, and Veritas can do it in software. (There is a Veritas for Linux, but it claims to be RH-specific.) All of those mean some complication on your networking concerns, though, and I do hope you don't plan on running anything *but* the synchronization across that gig line. -- gabriel rosenkoetter gr@eclipsed.net Attachment:
pgpP8AFoOMITn.pgp
|
|