I've been looking into this behaviour for quite a bit, I'm not sure if it's normal but strange or an issue, I'm hoping someone can provide some info:
I have a small linux server, but am using windows 7 as client. I have installed "Services for network filesystem" and was comparing SMB and NFS performance.
But I'm seeing strange performance. I have a SMB and NFS mount on the windows 7 machine. When I copy from the SMB mount to local machine, my copy-program says 7.5 MB/sec If I look in my windows 7 resource monitor I see 7.600.000 B/sec Receive
When i copy from the NFS mount to local machine, my copy-program says 5.5 MB/sec But when I look in my resource monitor I see 11.000.000 B/sec
So when I'm using NFS it's using double the bandwith than i'm actualy writing to my disk.
What is going on here?
On the linux machine under etc/export I have: /nfs/Gotjee 192.168.0.0/24(async,ro,no_root_squash,insecure,no_subtree_check,anonuid=0,anongid=0)
Any help or things I can test is appreciated, because the 11MB/sec in the monitor looks promising. And the main reason I want to find out what happening is because NFS is using much less CPU.
I am using NFS v3.
NFS:
150mbps wireless 27% util in Windows Task Manager/Networking 4.8MB/s in teracopy receive in resourcemonitor : 10.000.000 B/sec (10MB/s) linuw sent : 38Mb/s very consistant (4.75MB/s)
SMB:
150mbps wireless 42% util in Windows Task Manager/Networking terracopy 7.3 MB/s receive in resourcemonitor : 7.335.000 (7.34 MB/s) linux sent rate: 40-55 Mb/s varying (5MB/s - 6.9MB/s)
So it looks to be just some strange behaviour of the resource monitor. I tried a lot of configurations on server & client now to get my NFS speed up but I can't get it close or above my SMB speed. I wonder what the bottleneck is.