Hi!
Are there any limitations on the amount of data that can be used in the chart of Sankey module? I'm using 90MB file of 200K rows and the dashboard can hardly build a graph even with limit 10 and can never do it with limit 15 and above, while table data is calculated ok. Chrome or FF web-page simply crashes.
And can you please tell more about $sankey.searchterms$
Thank you!
p.s. I've asked this question in another one (d3-sankey-chart-without-using-django-framework), hope separately it will have more chances to be answred.
The SankeyChart module is a very early prototype I'm afraid, and there are quite horrible performance problems in it, such that it really hits a wall of death on relatively small datasets.
It seemed clear to me that some significant amount of this problem was from the core code itself (that I borrowed from other open source examples), but I can't say for sure that my modifications are not also doing something dumb. =/
But yes. You're not crazy. possibly it's some fundamental limitation on the algorithm that it uses but I think there's a nasty piece of implementation in there that scales badly, and it's waiting for someone to find it and fix it.
I'm very curious if you find any answers elsewhere. I have use cases for the SankeyChart module and this performance problem is far and away the largest problem blocking its general release as a part of Sideview Utils.
Thanks for the answer. Then my next steps are to try this app http://apps.splunk.com/app/1772/ that I knew about from this post: http://i-splunk.blogspot.ru/2014/03/web-path-analytic-with-splunk.html
Or implement Sankey by our own like it is described here http://dev.splunk.com/view/webframework-splunkjsstack/SP-CAAAEN6
Maybe you can address the authors of these posts as probably they have also faced the same problem.
Looking forward for your news!