Is it? Google search indicates that it is "sometimes" a problem (with small packets). When CS is the SSL client, the TCP_NO_DELAY should be set by the server (in this case hypervisors). But the nature of the interaction is not high bandwidth nor latency sensitive, so it shouldn't matter. When it is the server, I guess we could set it, but again, nobody has complained.
On 1/9/14 10:58 AM, "Demetrius Tsitrelis" <dtsitre...@live.com> wrote: >In CloudStack's various uses of SSL, I never see that we disable Nagle. >Isn't this a performance killer?
