Thanks Jan! I suspected that the slower performance from CouchDB was caused by a combination of HTTP latency and frequent disk flushes. Your articles help give me the big picture of those tradeoffs. I will definitely emphasize CouchDB's fault tolerance and distributability tonight.
On Tue, Sep 1, 2009 at 9:07 AM, Jan Lehnardt<[email protected]> wrote: > Cool, good luck! :) > > If you need some fuel on the number thing: > > http://jan.prima.de/~jan/plok/archives/175-Benchmarks-You-are-Doing-it-Wrong.html > http://jan.prima.de/~jan/plok/archives/176-Caveats-of-Evaluating-Databases.html > > Cheers > Jan > -- > > On 1 Sep 2009, at 17:50, Jesse Hallett wrote: > >> I will be giving a presentation on CouchDB tonight at the Portland Ruby >> Brigade Meeting, which runs from 7-9pm at Robert Half Technology, 222 SW >> Columbia St, Portland, OR <http://calagator.org/events/1250457540>. If >> there are any folks on the list who are in Portland and are interested, I >> would be thrilled to see you there. >> >> The focus of the talk will be on high-level Ruby interfaces to CouchDB. >> But >> there will be a lot of people at the meeting who don't know much about >> CouchDB or why they would want to use it. So I will also give some >> background and explain what sort of things CouchDB is good at. >> >> At the last meeting there was another presentation comparing the >> performances of several new database systems with benchmarks. CouchDB did >> very poorly in these benchmarks, compared to Tokyo Tyrant, MongoDB, and >> the >> usual SQL implemenations, for every operation except for retrieving >> documents in bulk. One of the challenges I will face will be explaining >> that there are applications where CouchDB is the best choice despite those >> numbers. > >
