Here are some benchmarks
System
4x 1TB Western Digital WD10EADS (raidz1 for the pool)
1x 160GB Samsung Spinpoint SP160 (for system)
4GB RAM DDR2 800
CPU Athlon X2 4200+
Mainboard Gigabyte GA-MA69VM-S2 (AMD 690V based chipset)
OpenSolaris 2008.11 snv_101b_rc2 X86
Real life benchmarking
Moving a 500 gigabyte large test-set within the same ZFS-Pool but between different file systems (so it degrades to normal copying) will reach about 82Mbytes per second.

Synthetic benchmarking
Network through output
This was estimated from Client to OpenSolaris Box whereas client is a Phenom 9750 running on Ubuntu 8.10 x64. The measurement has taken place using netcat on both sides. Filesystem is not involved, so this value should mark the upper limit.
server
nc -l -p1234 > /dev/null
client
dd if=/dev/zero bs=32768k count=10 | nc 192.168.0.103 1234 10+0 Datensätze ein 10+0 Datensätze aus 335544320 Bytes (336 MB) kopiert, 4,14416 s, 81,0 MB/s
So you can see, that the onboard solutions are not very efficient. Both chipsets are just some cheap realteks.
to be continued...