To further incentivize folks to try BRK, the latest release comes with yet another disk usage optimization.

You can now choose between eager and lazy computation.

The eager mode will compute everything and store it on disk which is the preferred method if you want speed.

The lazy mode will, on the other hand, compute data each time it's requested.

This only applies to datasets by transaction, type, input and output.

Datasets by height or higher (date, week, ...) are eager as computing them on the fly would be very costly with little to no benefit.

Right now computed datasets take 200+GB when eager and 5+GB when lazy, so quite the difference and this is in raw mode without compression.

Some numbers when fetching 10 000 values:

Eager:

txindex_to_base_size: 324.625µs

Lazy (2 dependencies, 1 level deep):

txindex_to_weight: 1.413ms

Lazy (4 dependencies, 3 level deep):

txindex_to_fee: 20.285ms (1st call), 8.968ms (2nd call)

So it's slower but not the end of the world it should be good enough for most use cases

#Bitcoin