Re: [ecf-dev] Java remoting protocol benchmarks

[Scott Lewis <slewis@xxxxxxxxxxxxxxxxx>]

Christian Campo wrote:
> Sure you are right as usual :-)

Ah, I wish it were true :)

RE: RMI...what's really being tested in these tests, I believe, is the 
serialization performance...as the tests were done on the localhost in 
two separate processes (i.e. with no network).  So the java 
serialization perf is what the RMI tests actually show.  Java 
serialization is used in several providers and isn't likely going anywhere.

I agree the Hessian performance looks good...although IMHO performance 
is only one aspect to choosing a serialization format.  One reason that 
we've taken the approach of having multiple providers (with different 
serialization means) is that this allows the use of whichever 
serialization is needed for given use cases (i.e. for interoperability 
or other requirements). 

Thanks,

Scott