BrowserUk,

Just a thought about my attempt about 'Bidirectional Lookup'.

A technique used with 'Big Data' databases is 'database sharding'. It is usually used to split a very large databases between multiple computers. But for your requirements, it could be fulfilled using an array of multiple scalars. Depending on the 'key/address' you could select where the start and end would be for each element of the array, so you could use 'index' on a sub-set of the data.

Obviously, the larger the scalar, the longer the 'index' function will take. To use 'sharding' you have to know a lot about your data, but it may improve the lookup results dramatically, which is what you seem to need to satisfy your requirements.

Just a thought!

Regards...Ed