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
We need to support batch publishes for a few use cases:
Clients want a very high ingest rate on streams and so might want to send batches of messages that would result in a single ACK
We want to support almost transaction-like related writes where several publishes to a stream are treated as a group and that groups messages are all committed or not with no chance for half commits.
Use case
For typed KV we might expand a hash/map into many keys inside a bucket. Imagine a USER record holding name, address etc. The address would be 4 or 5 keys and when the address change all 5 keys MUST update or none of the 5 should update.
A batch feature will facilitate that.
Additional to the high throughput publish mode already mentioned, we should aim for orders of magnitude increase in write performance.
The text was updated successfully, but these errors were encountered:
Proposed change
We need to support batch publishes for a few use cases:
Use case
For typed KV we might expand a hash/map into many keys inside a bucket. Imagine a USER record holding name, address etc. The address would be 4 or 5 keys and when the address change all 5 keys MUST update or none of the 5 should update.
A batch feature will facilitate that.
Additional to the high throughput publish mode already mentioned, we should aim for orders of magnitude increase in write performance.
The text was updated successfully, but these errors were encountered: