incubator-cassandra-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Michael Kjellman <>
Subject Re: CQL or Thrift ?
Date Tue, 06 Aug 2013 05:54:50 GMT
Why? Please give me 3 reasons why this is good advice to the OP.

The frame based binary protocol will be 100% easier to implement in C than
the never-been-out-of-alpha libc Thrift implementation. I¹ve worked with
that code for a while, it¹s *not* pretty.

I can¹t think of one reason, other than ³I already have legacy code
written around it² Thrift is better than the Native Protocol. So please
backup your claim before spreading false information to the community.

On 8/5/13, 10:46 PM, "Aswani Kumar Vonteddu" <>

>I'd recommend Thrift.
>> Date: Mon, 5 Aug 2013 14:08:52 -0500
>> Subject: CQL or Thrift ?
>> From:
>> To:
>> Hi,
>> I need a client in C (not C++) to work with Cassandra, so since there
>> is no one yet I would do my own. So far I have checked, I can do it
>> through Thrift RPC port, or through CQL port. As I understand, CQL
>> doesn't support direct "mutate" or "get_range_slices" calls like
>> Thrift does , so being Thrift API more low level, it should be faster
>> for specific queries my application would execute. The question is,
>> what do you recommend me to use, Thrift or CQL3 native protocol?
>> Thrift API supports  CQL , but not the other way around. If commands
>> go through CQL, there will be some Java code executed and I don't know
>> which would be faster, sending the raw "mutate"s through thrift by C
>> application or executing a prepared CQL3 query by Java? I am
>> interested more in performance, rather than in easiness of use. Will
>> appreciate very much your comments.
>> Nulik

View raw message