linuxperformancetaptcp
Linux.
Is there a performance loss from using a tap device vs a hardware one like eth0 ?
The goal is to create a user space tcp/ip stack and avoid the kernel as much as possible.
I would like to be sure that developping on top of a tap device is the right thing to do.
Thanks
I'd recommend against options IPFIREWALL_DEFAULT_TO_ACCEPT. The default is to Default to Deny. The firewall comes up with just one rule deny ip from any to any and stays that way until a script configures exactly what traffic should get through.
Follow-Up Note: RSA (one of the world's leading security technology companies) was hacked recently when part of their firewall was disabled during a maintenance window. This really underscores how quickly a system can be compromised given the right conditions.
If you insist on disabling the firewall until you explicitly block unwanted traffic, please consider using the sysctl available by adding net.inet.ip.fw.default_to_accept=1 to loader.conf. This has the added benefit of being easily modified (no recompiling the kernel) if you change your mind at some point in the future.
(moved from the question itself to a separate answer instead)
According to a user on HN (a KVM developer?) this is due to small packet performance in Xen and also KVM. It's a known problem with virtualization and according to him, VMWare's ESX handles this much better. He also noted that KVM are bringing some new features designed alleviate this (original post).
This info is a bit discouraging if it's correct. Either way, I'll try the steps below until some Xen guru comes along with a definitive answer :)
Iain Kay from the xen-users mailing list compiled this graph:
Notice the TCP_CRR bars, compare "2.6.18-239.9.1.el5" vs "2.6.39 (with Xen 4.1.0)".
Submit this issue to a Xen-specific mailing list and the xensource's bugzilla as suggested by syneticon-dj
A message was posted to the xen-user list, awaiting reply.
Create a simple pathological, application-level test case and publish it.
A test server with instructions have been created and published to GitHub. With this you should be able to see a more real-world use-case compared to netperf.
Try a 32-bit PV Xen guest instance, as 64-bit might be causing more overhead in Xen. Someone mentioned this on HN. Did not make a difference.
Try enabling net.ipv4.tcp_syncookies in sysctl.conf as suggested by abofh on HN. This apparently might improve performance since the handshake would occur in the kernel. I had no luck with this.
Increase the backlog from 1024 to something much higher, also suggested by abofh on HN. This could also help since guest could potentially accept() more connections during it's execution slice given by dom0 (the host).
Double-check that conntrack is disabled on all machines as it can halve the accept rate (suggested by deubeulyou). Yes, it was disabled in all tests.
Check for "listen queue overflow and syncache buckets overflow in netstat -s" (suggested by mike_esspe on HN).
Split the interrupt handling among multiple cores (RPS/RFS I tried enabling earlier are supposed to do this, but could be worth trying again). Suggested by adamt at HN.
Turning off TCP segmentation offload and scatter/gather acceleration as suggested by Matt Bailey. (Not possible on EC2 or similar VPS hosts)
Best Answer
I'm posting this as an answer due to insufficient commenting permissions.
From my experience VPN over tcp - tap was abysmal at best over wireless connections due to the the nature of TCP. When switching to UDP no more connection timeouts occurred - even better with tun device. I would suggest you test under poor network connections both TCP and UDP and tap/tun.