A good example for what I understand in using Hector / pycassa / etc. is, if you wanted to implement connection pooling, you would have to craft your own solution, versus implementing the solution that is tested and ready to go, provided by Hector. Thrift doesn't provide native connection pooling ...?
There are a few scenarios / examples where using a library that abstracts the Thrift bindings will make your life easier ... and they are maintained and up to date generally in alignment with new releases of Cassandra. That's a +1 for me ... Nothing stops you from using Thrift .. depends on how much work you want to implement yourself. -sd On Fri, Jun 17, 2011 at 5:30 PM, Markus Wiesenbacher | Codefreun.de <m...@codefreun.de> wrote: > One question regarding point 2: Why should we always use Hector, Thrift is > not that bad?