Re: Query cost
From: Alexander Langer <alexander.langer_at_address.invalid>
Date: Mon, 11 Nov 2013 18:31:09 +0100
Message-ID: <l5r48s$hnu$1_at_solani.org>
Date: Mon, 11 Nov 2013 18:31:09 +0100
Message-ID: <l5r48s$hnu$1_at_solani.org>
> yes it is projection (sorry about the font!)
> I just wanted a simple approach; like for join br+br*bs using block nested loop join (b=number of blocks)
> Does it use sort merge or hash functions cause there is set difference?
I suggest to define the runtime of the operations Π, -, x.
Think about the data set and about the operands / arguments. For ΠR−S we should have ((rn*rb)-(sn*nb)) costs.
So what about the - operator ?
If you find an known algorithm for that you can use his runtime.
Received on Mon Nov 11 2013 - 18:31:09 CET
