Thanks again Rebert.
I just upgraded to 2.0.9 from 1.2.5 and had to make some small changes regarding Order By and IN used together. I had to disable paging. The upgrade was easy and worked well so far and hoping if I upgrade to 2.1.x when released I do not have to do any more code changes. I want to upgrade to 2.1 for the 50% performance improvement and hopefully improved stability. Regards, -Tony On Wednesday, August 20, 2014 3:15 PM, Robert Stupp <[email protected]> wrote: 50% is the official statement. It depends on the workload and machine capabilities. But AFAIK 50% is a rough estimate. CQL over the native protocol is generally faster than Thrift. Native protocol also allows the client (e.g. the DataStax Drivers) to get informed about cluster changes (node add/remove/up/down), schema changes, etc. It also allows asynchronous execution of commands and requires just a few TCP connections thus lowering OS resource usage. Am 20.08.2014 um 23:09 schrieb Tony Anecito <[email protected]>: One last question. I read on someones blog that 2.1 will be up to 50% faster than 2.0.9 and that CQL will be the fastest to use rather than thrift. Any comments/thoughts on that? >
