benchmarks

[megauspexx]

No description provided.

[megauspexx]

Hello, i was interested in this extension. I would like to use it to calculate rank like this

select count(id) as count from data where (rating_sum > ? or (rating_sum = ? and p_id < ?) ) <more filters> .

What do you think about performance redisql vs mysql for this query?

Will redisql give more performance?
I tried using demo.redisql.com but for now connection refused

[siscia]

Hi there,

yeah unfortunately I had to switch off the demo machine.

For your particular query, the engine doesn't really matter much.
As long as you have the correct indexes in place.

A composite index on rating_sum and p_id would already makes wonder for this particular query. Maybe it would be possible to find an even better index give the other filters.

The advantage of RediSQL is that you run it in memory. And you don't hit the disk. Then if your disk is fast enough this is not going to be a big issue.

Suggestion it is to try it and see if it goes fast enough for your use case.

Note of allert, this project is being discontinued in favor of zeesql, which is the version 2 of RediSQL.

More info on https://zeesql.com

What you can try is to start zeesql with docker, populate it with some data and see the performance. Something like:

$ docker run -d --name zeesql --rm redbeardlab/zeesql
$ docker exec -it zeesql redis-cli
> ZEESQL.CREATE_DB DB
> ZEESQL.EXEC DB COMMAND "create table data(rating_sum INT, p_id INT);"
> ZEESQL.EXEC DB COMMAND "INSERT INTO data VALUES (1,2),(2,3),(3,4);"
> ZEESQL.EXEC DB COMMAND "select count(id) as count from data where (rating_sum > ? or (rating_sum = ? and p_id < ?) ) <more filters>"

Of course, you will need more data to have reasonable numbers to compare!