Re: Backup and trimming data idea

[newbie Shai]

Dear Rick,
Can I say the rule as look for all the where key and the order by combine them as one index even how many fields it could be. But some times one query may require just one of the key for the where statement and another query may require more then one field to look for. So in this case can I have index(a) then index(a,b). How about join statement where the where keys come from different tables?

I am not clear for this.

> index(a,b) or index(b,a) is there any difference
Think about fields last_name and first_name. Let's say you have
INDEX(last_name, first_name)
like in a Phone Directory. How, let's say you want
WHERE last_name LIKE 'J%' and first_name = 'Rick'
That index would be useless, but this would be useful:
INDEX(first_name, last_name)

My question here why is this index(last_name,first_name) useless when we are looking it via the like 'J%'? Kind of confuse here can you give me details for this?


Regarding your link which you share I read about chunk deletes but there is a risk of table locking right? So what best option am I left then only I guess partitioning but my scenario is a link of tables. I was thinking worse case I partition each tables separately based on date then drop them separately? What is your idea on this?