<warning alert="This is a loooong question">
I would be interested in comments from readers who are using, or have tried to use, point-in-time recovery. I have not used it in my environment, and am having trouble seeing how it could be made to work in a large, busy database.
i.e. Let's say I've got a large, integrated inventory - purchasing system. I've got clerks in different departments entering requisitions, tenders, quotes, purchase orders, receipts, warehouse orders, material issues, returns-to-shelf, returns-to-vendor, stock-count adjustments, employee purchases-from-stock, price adjustments, on and on and on.
In this case the users entering all these transactions into the database are working from different sites 50 miles apart.
So, if I wanted to restore to, say, 1:44pm this afternoon (or yesterday afternoon), how could I figure out which of all of these transactions had been entered precisely before or after 1:44? How would I contact all the people involved and ask them to figure out where they were at that exact point in time?
In other words, it sounds almost unworkable to me. Have I got that wrong? Can point-in-time actually be practical in a busy (but not abnormally busy) environment like this? Or, as I suspect, do people not bother with it?
Any comments much appreciated.
</warning>
bperry
I would be interested in comments from readers who are using, or have tried to use, point-in-time recovery. I have not used it in my environment, and am having trouble seeing how it could be made to work in a large, busy database.
i.e. Let's say I've got a large, integrated inventory - purchasing system. I've got clerks in different departments entering requisitions, tenders, quotes, purchase orders, receipts, warehouse orders, material issues, returns-to-shelf, returns-to-vendor, stock-count adjustments, employee purchases-from-stock, price adjustments, on and on and on.
In this case the users entering all these transactions into the database are working from different sites 50 miles apart.
So, if I wanted to restore to, say, 1:44pm this afternoon (or yesterday afternoon), how could I figure out which of all of these transactions had been entered precisely before or after 1:44? How would I contact all the people involved and ask them to figure out where they were at that exact point in time?
In other words, it sounds almost unworkable to me. Have I got that wrong? Can point-in-time actually be practical in a busy (but not abnormally busy) environment like this? Or, as I suspect, do people not bother with it?
Any comments much appreciated.
</warning>
bperry
but i persoanlly get a business descsion around this as i personallyback the tranasctions ever 25 mins![[noevil]](/data/assets/smilies/noevil.gif "[noevil] [noevil]")
