Roman, I would start with figuring out exactly how much performance you’re losing because of the language and whether there is lower hanging fruit. For the record, I *hate* java, but my experience in performance work has taught me that with some exceptions, it’s rarely the language where the performance sink is.
Do you have any data that points to Java as being a performance problem? Cheers, -Oscar On Dec 19, 2013, at 11:46 AM, Roman Vasilyev <rvasil...@netflix.com> wrote: > Never will say that Erlang will be faster and memory consumption will be > lower than native code. Rest of it totally agree with Brandon, in a few words > I'm talking about serious parallel project on improving Cassandra basics. > > ------ Original Message ------ > From: "Andy Cobley" <acob...@computing.dundee.ac.uk> > To: "<dev@cassandra.apache.org>" <dev@cassandra.apache.org>; "Roman Vasilyev" > <rvasil...@netflix.com> > Sent: 12/19/2013 11:40:55 AM > Subject: Re: C* engine > >> The question just as easily be asked, why not rewrite it in Erlang ? >> Probably lots of advantages to be gained there, the ability to upgrade on >> the fly, native message passing, functional programming paradigm ? >> >> But as Brandon says, you would be throwing away so much work, would it be >> worth it ? >> >> Andy >> >> >> On 19 Dec 2013, at 19:22, Roman Vasilyev <rvasil...@netflix.com> wrote: >> >>> Hello, >>> >>> Don't want to rise "holy war". Just let me share my crazy thoughts. >>> I believe it could improve Cassandra speed and robustness. >>> >>> What people will say if I propose to have Cassandra engine written in >>> C/C++, and this engine will give you ability to run extensions in Java, >>> Groovy and bunch other languages like Perl/Python/Ruby? >>> >>> I just want to understand for myself does this solution will be useful or >>> I'm looking in wrong direction? >>> >>> Thank you for reading. >> >> >> The University of Dundee is a Scottish Registered Charity, No. SC015096. >> >> >
