sql - What indexing implemetations can handle arbitrary column combinations? -

I am developing a little data warehouse system with a web interface where people can now filter filtered searches There are about 50 columns that people might want to filter, and about 2.5 million lines. A table scan pain is slow. The problem is that the queries that I'm getting are not having any normal prefix.

I am currently using sqlite3, which will only use one index if the column is needed, then the left column in that index. This means that I need many indexes, a quick look at MySQL shows that many types of indexes will also be required for this type of query.

My question is whether indexed implementations are available for different database systems, which can handle such queries on the arbitrary combinations of columns?

I have prototyped my indexing scheme; I am collecting additional tables that round the primary key in my big table, where each value is for each column, and I have enough data to check prices with the lowest number of matches first. It works fine; The table is better than the scan, but a little bit slow, which is unsafe for the first version in Python, many SQL queries are doing.

There is store data on a per-column basis, where each column is its index They are very good fit for the data warehouse, because they are fast enough to read, but they are quite slow to update.

An example is an optimized MySQL engine, and for a number of top crowning weeks, at an impressive system cost, note that kick-fire is a device that is sold as a hardware box. She goes.

There will be another example, and there is a free in it that runs on Windows and Linux.