One or two indexes? add_index :votes, [:item_id, :ip_address]

What’s the difference between

add_index :votes, [:item_id, :ip_address]

and

add_index :votes, :item_id
add_index :votes, :ip_address

Not just in rails, but in the db as well?

Thx!

For the first you’re adding the index on the item_id AND ip_address
fields,
together.

For the second you’re adding them on the fields individually.

On Jan 18, 2008 5:07 PM, Stedwick [email protected] wrote:

Not just in rails, but in the db as well?

Thx!


Ryan B.

Feel free to add me to MSN and/or GTalk as this email.

So, like, when would I use one or the other? What’s the benefit of
having an index on two columns, for example?

Sorry, I guess this is more an SQL question than a rails question.

On 18 Jan 2008, at 07:32, Stedwick wrote:

So, like, when would I use one or the other? What’s the benefit of
having an index on two columns, for example?

Sorry, I guess this is more an SQL question than a rails question.

I can only tell you about MySQL, but as far as it goes there’s a big
difference.
MySQL will only ever use one index per table, so if you do

Votes.find_by_item_id_and_ip_address
with 2 separate indices then mysql will pick the one it things is the
best (most selective) and ignore the other one. With a combined index
it will find things faster.
Note that if you have an index on [a,b,c,d] then that index can be
used as an index for any left prefix (ie it can be used as an index on
a, [a,b] but not on b alone.
It will also help you with things like Votes.find :all, :conditions =>
[“item_id = 123 AND created_at >= ? and created_at < ?”] (if you had a
combined index on item_id and created_at). Similarly it will help
mysql if you did a select of all Votes with a give item_id and then
sorted by ip_address.

So a combined index can really help things out. The problem is that
since order matters, there are a lot of possible combined indices: if
you had 5 columns you were interested in, then you could have 5! (120)
different combined indices which would probably be a bad idea, so you
have to pick the ones that matter.

Fred

Excellent description. I got it now. Thx!

On Jan 18, 4:20 am, Frederick C. [email protected]

Using:
add_index :votes, [:item_id, :ip_address]

Will create a single index on both columns, item_id first.

This will create an index with higher selectivity and is a better index
ONLY IF you always use both columns in a select query. With most
database engines you the index will also effectively work if you just
use item_id on its own in a select, but NOT using just ip_address.

Using:
add_index :votes, :item_id
add_index :votes, :ip_address

Will create two separate indexes and both may be used on their own in a
select query. Depending on the database engine it may be possible that
both are used in a select that uses both in the where clause. Combining
two separate indexes for a single query is highly dependent on the
database engine you are using, so check the documentation for your
database. Postgresql and MSSQL server can do it, not sure about MySql.

Cheers,
Gary.

On Jan 18, 1:20 am, Frederick C. [email protected]
wrote:

So a combined index can really help things out. The problem is that
since order matters, there are a lot of possible combined indices: if
you had 5 columns you were interested in, then you could have 5! (120)
different combined indices which would probably be a bad idea, so you
have to pick the ones that matter.

Another use of a multi-column index is the idea of a “covering” index,
which contains all the columns in a particular query. Only one or two
of the index columns are actually used for querying - the other
columns come along for the ride so that, having found the right index
entry, there’s no need to go back to the table.

///ark