On Thu, May 5, 2011 at 11:21 AM, Rajesh Kumar G <[email protected]> wrote: > Hi Jesse, > > I was trying to see if the performance degrade that is observed with > GRE+IPSec as mentioned in this mail thread > > http://www.mail-archive.com/[email protected]/msg00915.html > > can be overcome by trying CAPWAP tunnel instead. Is there any fix for the > GRE+IPSec performance degrade yet? I observed the degrade in > openvswitch-1.1.0 released code also.
IPsec is the expensive operation here, not the tunneling. You won't see any substantial difference between GRE and CAPWAP. This has more to do with the Linux kernel than OVS. We have seen some improved results with 2.6.38. In addition, your choice of processor and cipher can make a large difference. The newest generation of Intel processors have support for AES instructions and the triple DES cipher that you have chosen is extremely slow. _______________________________________________ discuss mailing list [email protected] http://openvswitch.org/mailman/listinfo/discuss
