totally agree. rgds, canal
________________________________ From: Chris Anderson <[email protected]> To: [email protected] Sent: Saturday, September 26, 2009 1:36:00 PM Subject: Re: Slide on CouchDB: very very very slow ? On Fri, Sep 25, 2009 at 8:19 PM, Paul Davis <[email protected]> wrote: > http://vmx.cx/cgi-bin/blog/index.cgi/benchmarking-is-not-easy%3A2009-09-23%3Aen%2CCouchDB%2CPython%2CTileCache%2Cgeo > http://jan.prima.de/plok/archives/175-Benchmarks-You-are-Doing-it-Wrong.html > http://jan.prima.de/plok/archives/176-Caveats-of-Evaluating-Databases.html > > HTH, > Paul Davis > > On Fri, Sep 25, 2009 at 11:00 PM, go canal <[email protected]> wrote: >> I just saw >> http://github.com/igal/ruby_datastores/raw/master/2009-08-04%20Non-relational%20data%20stores%20for%20Ruby%20r1.pdf >> >> the comments for CouchDB is 'very very very slow'... I am not sure about the >> context. >> >> >> Slide #13 has some performance data. The difference is significant. >> >> any comments ? >> rgds, >> canal >> >> >> >> > There is an emphasis placed on serial speed in benchmark-land. Serial speed is important if you are doing big imports from a handful of clients. There are precious few benchmarks of these databases that compare how many concurrent connections they can support, which is a much more important metric for operating at scale, than raw query execution speed. What matter is fast enough, and not falling over under load. Chris -- Chris Anderson http://jchrisa.net http://couch.io
