ts, we would prefer
> setting up machines with spinning disks (2TB per node) to save costs.
>
> If you have spinning disks and 1G networking the rule of thumb is 300GB to
> 500GB per node. See previous discussions about size per node.
>
> Cheers
>
> -----
>
Hi all,
We are working on a data model for storing tweets for multiple streams
(where a stream is defined by a number of keyword filters on the full
twitter firehose), and retrieving the tweets by timestamp and hashtag. My
question is whether the following data model would a good way for doing
tha
