You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently the tree has a lazy root by default, which reduces contention when modifications are random. But if modifications are always at the end of the tree, or clustered in a single subtree, we are back to serializing all writes.
To fix this, we should make inner nodes become lazy dynamically when they are being written too quickly. Something like this:
Currently the tree has a lazy root by default, which reduces contention when modifications are random. But if modifications are always at the end of the tree, or clustered in a single subtree, we are back to serializing all writes.
To fix this, we should make inner nodes become lazy dynamically when they are being written too quickly. Something like this:
https://stack.convex.dev/rate-limiting
And then if the node isn't written to for a while, we can make it non-lazy again.
Potentially we would want a cron to make it non-lazy in case there are a ton of writes all at once and then a read-heavy workload afterwards.
The text was updated successfully, but these errors were encountered: