Hello,
I'm a bit concerned by network performance on child partitions with Hyper-V. Here is my test environment :
- One physical server with two physical NIC (one dedicated to parent partition, the other one for child partition)
- Windows Server 2012 OS on parent & child
- External virtual network bound to second physical NIC (host OS
not allowed to share this network adapter)
The problem is as follows : from a random device on network, ping response time of host is excellent (always < 1ms), but is *a little* worse for guest (on average< 1ms, but sometimes higher, = 1ms, 10ms or even > 30ms). To be more clear, here is ping results for host :
Pinging 192.168.1.8 with 32 bytes of data:
Reply from 192.168.1.8: bytes=32 time<1ms TTL=128
Reply from 192.168.1.8: bytes=32 time<1ms TTL=128
Reply from 192.168.1.8: bytes=32 time<1ms TTL=128
Reply from 192.168.1.8: bytes=32 time<1ms TTL=128
Reply from 192.168.1.8: bytes=32 time<1ms TTL=128
Reply from 192.168.1.8: bytes=32 time<1ms TTL=128
Reply from 192.168.1.8: bytes=32 time<1ms TTL=128
Reply from 192.168.1.8: bytes=32 time<1ms TTL=128
Reply from 192.168.1.8: bytes=32 time<1ms TTL=128
Reply from 192.168.1.8: bytes=32 time<1ms TTL=128
Ping statistics for 192.168.1.8:
Packets: Sent = 10, Received = 10, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 0ms, Maximum = 0ms, Average = 0ms
For guest :
Pinging 192.168.1.9 with 32 bytes of data:
Reply from 192.168.1.9: bytes=32 time=1ms TTL=128
Reply from 192.168.1.9: bytes=32 time=3ms TTL=128
Reply from 192.168.1.9: bytes=32 time<1ms TTL=128
Reply from 192.168.1.9: bytes=32 time<1ms TTL=128
Reply from 192.168.1.9: bytes=32 time<1ms TTL=128
Reply from 192.168.1.9: bytes=32 time<1ms TTL=128
Reply from 192.168.1.9: bytes=32 time=1ms TTL=128
Reply from 192.168.1.9: bytes=32 time<1ms TTL=128
Reply from 192.168.1.9: bytes=32 time=5ms TTL=128
Reply from 192.168.1.9: bytes=32 time<1ms TTL=128
Ping statistics for 192.168.1.9:
Packets: Sent = 10, Received = 10, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 0ms, Maximum = 5ms, Average = 1ms
As you can see, guest results are just a little bit worse than host (I repeated this test dozens of times from different devices on network, results show the same pattern each time).
I guess the problem is due to the fact that parent partition can access directly its dedicated NIC, while the child partition must pass by the virtual switch (which is IMO completly useless in this case). How can I solve this problem ?