Hector vs Astyanax for Cassandra [closed]

Question

We are starting a new java web-project with Cassandra as the database. The team is very well-experienced with RDBMS/JPA/Hibernate/Spring but very new to the world of NoSQL. We want to start the development with as simple setup as possible. Hector seems to be the most preferred and popular choice for connecting to Cassandra. But, Netflix has recently offered Astyanax, which has its origins in Hector. Can anyone who has used both these technologies share their experiences? I am looking for easy setup, good documentation and simple/clean usage. Suggestions about other api's are also welcome.

Answer 1

I've tried both and Astyanax is way easier. The API actually makes sense and reflects what your are actually doing. Both Hector or direct Thrift usually results hard to decipher code.

There are some issues yet to be solved in Astyanax (a.o. getColumnByName), but I've decided to build my project using it.

Oh, I used the snapshot version (manually build, since it was not in any maven repo) because of some outdated references.

Answer 2

FWIW, I've only been working with Cassandra for about 2 weeks now, but I'm already successfully writing to Cassandra with Astyanax, using composite columns, etc.

I am also not a Java programmer either, so my point is that it seems pretty robust and easy to use.

Answer 3

I started using Hector, which I think is excellent, but then moved to Astyanax. Astyanax is an abstraction over Hector, and is much simpler to use IMHO. I experienced a minor bug on Hector, to do with keeping a connection alive, which was magically solved once I'd switched to Astyanax.