[
https://issues.apache.org/jira/browse/FLINK-14147?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16955713#comment-16955713
]
Jiayi Liao commented on FLINK-14147:
------------------------------------
This should be a great improvement. Curretly we suffer the slow reponding of
Flink UI a lot when running high-parallelism jobs.
> Reduce REST API Request/Response redundancy
> --------------------------------------------
>
> Key: FLINK-14147
> URL: https://issues.apache.org/jira/browse/FLINK-14147
> Project: Flink
> Issue Type: Improvement
> Components: Runtime / Web Frontend
> Reporter: Yadong Xie
> Priority: Major
> Attachments: 16_17_07__09_20_2019.jpg
>
>
> 1. Redundancy Response
> In the response of API /jobs/:jobid, the id and name in both plan and the
> vertices data are exactly the same, it would waste a lot of network bandwidth
> if the vertex graph is very big(1000+ vertex in a job).
> !16_17_07__09_20_2019.jpg|width=427,height=279!
> 2. Redundancy Request
> In the current Web UI design, we have to query vertex number times to display
> the low watermarks in the job graph. If the vertex number is very large(1000+
> sometimes), the Web UI will send 1000+ request to the REST API, the max
> concurrent HTTP request in a browser is limited, it would bring a long time
> delay for users.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
