[TYPO3-dev] Testing TYPO3 caching vs.[FE][pageCacheToExternalFiles]

[Martin Kutschker]

Dmitry Dulepov schrieb:
> --------
> When you precede a SELECT statement with the keyword EXPLAIN, MySQL
> displays information from the optimizer about the query execution
> _plan_. That is, MySQL explains how it _would_ process the SELECT,
> including information about how tables are joined and in which order.
> --------

It's "would" and not "will" because no real select takes place. If you're 
doing an EXPLAIN followed by a SELECT with the same query, Mysql WILL do 
what it said it WOULD.

>>Adding indices without checking EXPLAIN before and after making changes
>>to indices is IMHO only wild guessing.
> 
> I agree. My point is that EXPLAIN is not to be trusted for 100% because
> it depends on the existing database data and it is difficult to build
> proper indexes unless there is real data in the database.

I have the feeling that we are basically agreeing, but having troubles to 
see that. I'm no native speaker and from your name I gues you aren't.

My point is that you cannot make good indices without data. You need a real 
world data to make indices. If you want to ship Mysql with good all-purpose 
indices you must collect some stats on average DB usage. IMHO it won't do 
any good to add indices just because it "feels right" to add some.

This is why I say no to adding indices to "deleted" and "hidden". If it can 
be proven that there is a performance beneift (where? BE? FE? page 
rendering? ...) then add it. If you cannot prove it, don't add it.

Masi