Question

I have a table called clicks which records mouse clicks around a website. The schema looks like this:

There are 3 main queries run on this data:

• “SELECT * WHERE pageid = xxxx”
• “SELECT * WHERE visitorid = xxx AND pageid = xxxx”
• “SELECT visitorid, MIN(time) AS mintime WHERE pageid = xxxx GROUP BY visitorid”

In the future I might use maintestid in queries too. So this would look like:

• “SELECT * WHERE maintestid = xxxx”

So I’ve added 2 indexes to the table for both these queries along with the primary key like so:

I have 3 questions:

1. Is this primary key useful at all being like this? Or should I make it the ID column only?

2. Is the pageid index useful at all? Or will it be just as fast if all queries for pageid are be handled by the pageid_visid index?

3. When I want to search via maintestid in the future, should I just add an index on the maintestid column, or something different?

Answer 1

1. The primary key is NOT useful like you have it – the other columns of that index will never be used. Just make it on the id column only
2. The pageid only index is slightly useful in that its entries will fit on less index pages, but I would drop it and let mysql use the other pageid-based index (do a benchmark test to confirm)
3. Yes – define your indexes to match your important queries