andrea This might interest you - I've just time tested my new dump routine, which doesn't write to file with every insert (but just one data dump per table) and the speed test for a 4.92Mb database is:


Based on average of 10 dumps per script (we need to be scientific about it, right ).

My new script is almost 9 times quicker, and should write your backup in (115/4.92)* 1.7 = 40s! (depends of course on servers... actually working backwards, 115/4.92*15.1 = 353s - so I guess your server is only a tad slower than mine).

One problem to address with this method is my multi-file spanning feature (check the current demo link above for this, though current demo does not include latest changes). Using a table dump variable to hold all the data means the string needs to be checked for size, and then split up depending on the max file size spec. That, however, shouldn't slow down too much, just yet another change in strategy - and if you don't set a max file size, you'd get the full speed improvement. And I might squeeze a little more performance out of it yet

Thanks moreteavicar, I read your PM (and I answer too). I'll test new version ASAP, and I will say if it will works fine or not. Thanks for now!

____________
My Icy forum: www.lineameteo.it
Staff Member @ IcyPhoenix.com & IcyPhoenix.it