Also keep in mind that not all deployments of Cassandra are performance-critical.
My current project is an integration/legacy MySQL replacement, so the database is used as a system of record but not of (significant) runtime. The read paths are fronted by caches with 97% hit rates. A JSON http API would be more than performant enough, and ease the integration path by not requiring a new component (Thrift) in the client stack. Having a quick time-to-production through easy integration is much more important to us. Then, if production experience suggests that we can remove some of the caches as a phase 2, we will worry about transport performance. Evan On Wed, Jun 24, 2009 at 9:09 PM, Brett Eisenberg<[email protected]> wrote: > +1 for Avro as something to watch and consider. I concur that performance > and simplicity are the quintessential attributes to prioritize; for me that > excludes code generation. > > +1 for something simple like HTTP+JSON or line-based as an interface of last > resort/debug. > > b > > On Jun 24, 2009, at 14:03 , Jonathan Ellis wrote: > >> I'm interested in Avro as kind of a better Thrift, but that's still >> pretty alpha. >> >> I'm not really interested in stuff that's going to be Much Slower like >> anything over http (Jay from Voldemort said that's basically a waste >> of time and I believe him) or anything that requires hand-writing >> clients for each language (everything else?). >> >> -Jonathan >> >> On Wed, Jun 24, 2009 at 3:53 PM, Evan Weaver<[email protected]> wrote: >>> >>> I wanted to start a small discussion to see if there is any interest >>> in supporting alternative wire protocols or perhaps junking Thrift to >>> some degree. >>> >>> Some options: >>> * Use JSON over HTTP >>> * Use BSON over...something (http://www.mongodb.org/display/DOCS/BSON) >>> * Use ASN.1 over...something >>> * Use Protocol Buffers over...something >>> * Use Thrift, but package Cassandra-specific clients for each language >>> >>> I have not thought too coherently about this but generic Thrift seems >>> to be a pain point for everybody. >>> >>> Evan >>> >>> -- >>> Evan Weaver >>> >> >> !DSPAM:1000,4a42938f192413113311398! >> > > -- Evan Weaver
