[
https://issues.apache.org/jira/browse/STORM-69?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14061280#comment-14061280
]
ASF GitHub Bot commented on STORM-69:
-------------------------------------
Github user knusbaum commented on the pull request:
https://github.com/apache/incubator-storm/pull/127#issuecomment-48962364
@Gvain
I had hoped that the stream selection checkboxes at the top would be enough
to mitigate this issue, allowing you to view only the portion of your topology
that you're interested in. Unfortunately, I don't know if it's really feasible
to visualize a gigantic topology in its entirety.
Two ideas on how to fix this:
- Your idea of a "drag and stick" mode to prevent the components from
bouncing off-screen
- In addition to the stream-selector, a component selector for
finer-grained control over exactly which components are part of the graph.
Please file a Jira for this and I'll give it a look.
> UI Visualizations
> -----------------
>
> Key: STORM-69
> URL: https://issues.apache.org/jira/browse/STORM-69
> Project: Apache Storm (Incubating)
> Issue Type: New Feature
> Reporter: James Xu
> Assignee: Kyle Nusbaum
> Fix For: 0.9.2-incubating
>
>
> https://github.com/nathanmarz/storm/issues/368
> I've worked with Storm for a while now, and I can say that having a quick
> visual on the makeup of your topologies is worth a lot. It's far easier to
> simply look at a graph, and see the connections between named nodes, than to
> browse through code. Invaluable when you're debugging a problem in production.
> Along these lines, I think it would be a real asset to have a handful of
> useful visualizations in the UI.
> One example would be circle packing, where the containing circles are
> topologies, and the inner circles are topology components. See
> http://mbostock.github.com/d3/talk/20111116/pack-hierarchy.html. The idea
> would be to write it flexibly, so that the size of each circle is given by a
> function over the topology and/or node. You could imagine representing such
> things as throughput, latency, (received - acked), (# of executors occupied),
> etc in this way. It would be a quick visual indicator of how your topologies
> are behaving.
> Another example would be if you wanted a quick visual depiction of a topology
> -- you might use a graph, or a force-directed graph:
> http://mbostock.github.com/d3/talk/20111116/force-collapsible.html
> Many more ideas here: https://github.com/mbostock/d3/wiki/Gallery
> If you guys think any of these would be a valuable addition, I'd love to take
> a crack at it.
> ----------
> nathanmarz: I would like to see a force-directed graph of each topology, with
> the size of the edges between nodes indicating the amount of throughput
> passing between them. Once we improve the metrics (notably issues #362 and
> #363) we can use color to indicate which components are likely bottlenecks in
> the topology.
> It would be really cool if clicking on a node shows more detailed information
> (like a visualization of the incoming and outgoing throughput to each
> individual task in that component).
> ----------
> gsilk: I'm on it.
> ----------
> chenxiaozhao: i can‘t wait to see the force-directed graph ,which can show
> the incoming and outgoing throughput to each individual task in that component
--
This message was sent by Atlassian JIRA
(v6.2#6252)
