hanahmily commented on issue #13811:
URL: https://github.com/apache/skywalking/issues/13811#issuecomment-4230398545
Neither "intra-node truncation" nor "cross-node truncation" is a real
problem for the current architecture, because entity-to-shard affinity
guarantees that all data for a given entity
is processed by a single node's streaming TopN processor. There's no
partial aggregation across nodes to worry about.
> Cross-node truncation: An entity that ranks low on one data node might
rank high globally after merging partial results across all nodes.
This would only be true if the same entity's data were split across multiple
nodes — for example, if COUNT aggregation required summing partial counts from
node A and node B. But since entity routing is deterministic by shard:
- Entity5's all data points land on exactly one node
- That node's streaming processor computes the correct TopN ranking for
entity5 based on all its data
- If entity5 doesn't make the local top-N, it genuinely has a lower value
than the N entities that did — no missing partial data exists on other nodes
> Intra-node truncation: A single data node may have multiple data points
for the same entity (e.g., across different shards),
It's not true. The same entity's data points always go to the same shard.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
