Ruslan Dautkhanov created ZEPPELIN-3251:
-------------------------------------------
Summary: slow rendering on very wide datasets - display columns
only when necessary
Key: ZEPPELIN-3251
URL: https://issues.apache.org/jira/browse/ZEPPELIN-3251
Project: Zeppelin
Issue Type: Improvement
Components: front-end
Affects Versions: 0.8.0, 0.7.4, 0.9.0
Reporter: Ruslan Dautkhanov
Apache Zeppelin recently upgraded to angular ui-grid - that looks awesome!
What we're facing though is on wider datasets Chrome sometimes spends minutes
to render a table.
The table doesn't have a lot of data (it's very sparsely populated), so it
seems just the number of columns is what important.
Maybe more of an extreme case - 5000 columns x 1000 rows recent Chrome spent
many minutes to render on a decent PC. That's not a make-up case and we
actually had to display such a wide sparsely populated table. (Our company
normally deals with wider datasets)
Even on not so wide datasets we found browser rendering could be lagging..
*Possible solution: shows rows/columns lazily on scrolling events*.
As an example, Cloudera HUE worked around displaying such wide datasets by
adding "virtual" horizontal scrolling - basically columns are pushed to an html
table only when they have to be visualized (if a user has pushed horizontal
scrolling to the right .. and efficiently removing columns from html when
they're not necessary for visualization). Cloudera Hue can render such wide
datasets very quickly because browser has to render only a tiny subset of
columns that it actually needs.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
