On Tue, May 26, 2015 at 12:21 PM, Vladimir Rodionov <[email protected]> wrote:
> One suggestion: > > Can we move from simplistic scenarios towards more realistic ones? When > client - server interaction can involve server-side > processing as well? HBase is good at that. > > One example: credit card fraud application which usually, for every > transaction, involves analyzing client TXs for past xx months as well > as verification of a vendor (simplistic description but nevertheless ...)? > > Another example: sensor data stream -> server calculates and updates > multiple aggregates for every sensor reading. > > -Vlad > > I'd love to see more realistic workloads represented. We'd have to come up with a reasonable abstraction that could be implemented by the different datastore projects. My guess is this will take some discussion. Care to open up an issue on the github repo[1]? [1]: https://github.com/brianfrankcooper/YCSB/issues -- Sean
