François Massot

> We would still have the struct with data_start_offset, since we shouldn't calculate it on every access. So the removal of data_start_offset would be only in the serialized data, not...

@fulmicoton linear interpolation works great on sorted data, on HTTP logs, the compression ratio is 0.26 for bitpacking, 0.05 for multilinear. I tested your idea with blocks of 128 elements...

I added the frame of reference implementation in #1217 to compare the results

This a bug in the current implementation, it has been addressed in #1217

@kstaken some short answers so that you know a bit what's going on there: > Is there a ticket that covers the design and reason for introducing the control plane...

Closing this old PR which is replaced by https://github.com/quickwit-oss/quickwit/pull/1853

I think this can be closed thanks to #1752 or did you have more in mind @fulmicoton ?

We are working on this, we will share more details about the design soon :) On the cluster formation, we already use [chitchat](https://github.com/quickwit-oss/chitchat) (based on anti-entropy gossip algorithm scuttlebutt). The...

> Are you going to store the routing information in meta store and let the indexing task read the information from meta store? Yes, the typical example is for a...

True, thanks for reporting that. I opened a PR here https://github.com/quickwit-oss/quickwit/pull/1696