When I talk about a benchmark (and expecially about a benchmark regarding
performance comparison between different platforms) I'm always sceptic about the
results. They could not be the reality, but sometimes they give some
signs...
Today I was reading a great Benchmark
comparison about Web Services published by TheServerSide.NET and I've see some
interesting things from the results given.
This test was taken by using WSTest
1.1, a Microsoft implementation of WSTest 1.0 that mirrors
the Sun implementation of the EchoVoid, EchoStruct and
EchoList tests (where messages of different complexity are exchanged)
and, in addition, it includes one new test called GetOrder that
tests a more complex object type that simulates a purchase order. The second
test was about XML Performances and was taken by
using XML Test, an XML processing test
developed at Sun Microsystems and designed to mimic the processing that takes
place in the lifecycle of an XML document (Parse, Access, Modify and Serialize).
If someone is sceptic, the tests used are open source and can be
replicated. 
Why the results are interesting? Not because they show that .NET Web Services
are in general more performant than Java WS (this could not be the reality on
every cases) but expecially because they show that .NET 2.0 Web Services are
more performant than .NET 1.1 Web Services (normally from 25% to 50% of
performance improvements).
This is a great sign and another reason for a migration to the new platform
(expecially if you plan to work with high performance Web Services).
These are some reports:
