[
https://issues.apache.org/jira/browse/FLINK-10742?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Zhijiang updated FLINK-10742:
-----------------------------
Release Note: After FLINK-10742, while reading data stream from network on
receiver side, it would reuse the Flink's buffers directly to avoid extra copy
from Netty's ByteBuf to Flink's buffer. Therefore the Netty internal memory
overhead is greatly reduced, which can solve the potential direct
OutOfMemoryError specially in large-scale jobs.
> Let Netty use Flink's buffers directly in credit-based mode
> -----------------------------------------------------------
>
> Key: FLINK-10742
> URL: https://issues.apache.org/jira/browse/FLINK-10742
> Project: Flink
> Issue Type: Sub-task
> Components: Runtime / Network
> Affects Versions: 1.7.0
> Reporter: Nico Kruber
> Assignee: Yun Gao
> Priority: Major
> Labels: pull-request-available
> Fix For: 1.11.0
>
> Time Spent: 20m
> Remaining Estimate: 0h
>
> For credit-based flow control, we always have buffers available for data that
> is sent to use. We could thus use them directly and not copy the network
> stream into Netty buffers first and then into our buffers.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
