How about changing the filtering conditions for your demo? On Wed, Nov 5, 2014 at 8:14 PM, Anirvan BASU <[hidden email]> wrote:
|
Yes, I'd also play around with the filters to get some output. Not sure what exactly you want to demonstrate, but processing a few MBs on a 10 node cluster might look a bit strange... How about switching to another example that works on less specific data? There are a nice graph processing examples and also publicly available graph data set. Cheers, Fabian 2014-11-05 21:14 GMT+01:00 Kostas Tzoumas <[hidden email]>:
|
Fabian and Kostas, Thanks for your suggestions. Fabian, I do agree with your point about processing a few MB of data with a 10-node cluster and a framework capable of processing several 100 GB of data. The reason why I am trying to do this is because of the dataset provided by one of the companies - not so interesting dataset :-(( However, I am open to changing the Use Case as per your suggestions and ideas. What specific graph processing examples would you suggest that can be done with Flink ? It helps if there is a complex underlying workflow .... (than just a simple WordCount with GB datasets) As Flink has a graphic interface (webclient) I would like (if possible) to use it in my demonstration - it helps to attract the audience. Another possibility that I was considering was to demonstrate the TF-IDF with some good dataset - any suggestions there ? Thanks in advance for all your advice, Anirvan From: "Fabian Hueske" <[hidden email]> |
Free forum by Nabble | Edit this page |