Re: Point in time recovery

Nils Myklebust wrote: > May be you could go to the Informix development department with this > utility and have them create a supported version? May be you ended up > working there as well :-) Why would I want to do that? What could possibly be better than the job I have now :) > May be one day it would support onbar as well? I'll look into it, just as soon as I finish addressing some of the other limitations (no support for fragmented tables, detached indexes, GLS, 64-bit OS's, ummm...) Don't wait for onarchive support; it's my goal to die without knowing substantially more about onarchive than I do now. > As it does an onunload may be one day it could do an unload (ascii > format) as well? That might be usefull in making it simple to check > the data before it's reloaded into the database. Yes, I'm thinking about that. But one advantage to the way it works now is that (in theory, at least) you can load the table into a test instance and oncheck it without locking up your production system. While it is (again, theoretically) possibly for the unload/load process to introduce index corruption (if there was activity on the table during the archive), as a general rule I wouldn't expect this to be the case, and in any event the unload/load will not FIX any corruption, so if the oncheck comes up clean on the test instance, you can be pretty sure it's okay on the source. I think this could be very useful for a 24x7 shop that doesn't have a window for doing oncheck's. Hmmm, I should probably mention that on the webpage. > I realize the nature of the utility and these are *not* requests of > you to do anything about them. This is just meant as comments of what > might be possible to do and may be the developers could one day do > something about it. That's fine; I'm open to suggestions. They'll just have to be prioritized along with all the other things it doesn't do right now, plus factored into my schedule. > Table level restore was after all quite high on > the IIUG wish list and it seems to me that this utility could solve it > quite elegantly. I think they have committed to providing some type of table-level restore in some future release (sorry for the vagueness). It probably won't be related to my utility, though.