Hey everyone, We’ve been doing lots of work recently to try and improve the speed of our requests as they are a little on the slow side unfortunately and we’ve become a bit stuck with numerical filtering. We have a price filter which could be any value from

0

to

100_000_000

and in our UI we use a range slider to allow the user to select what prices they would like to see. We then form a filter like so

filter_by: price:[100..500000]

as an example. We have about 3.5 million records and on average this search takes 1 second to complete. Our current ideas are to try all of the following: • Adding

ranged_index

to the schema ◦ We’ve done this for part of our data set and so far we are only seeing a very small improvement • Increase RAM • Move to run in a VM • Move to Typesense cloud and if all of that doesn’t help, then we have thought about trying to turn it into a string based filter with a set of ranges, so something priced at

10

would be assigned

priceFilter: '0'

(range 0 to 100) and something priced at

12000

would get

priceFilter: '10000'

(range 10_000 to 20_000). We would then turn this into a filter like so

filter_by: (priceFilter:='0' || priceFilter:='100' || priceFilter:='1000' || priceFilter:='10000')

We would like to avoid doing this though as it would be a huge undertaking to apply this new

priceFilter

to all 3.5 million records. If there is a better way or something that we are missing I would really appreciate some advice. Thanks in advance.