DX Infrastructure Manager

Expand all | Collapse all

Question regarding performance /bandwith using SSL tunnels..

  • 1.  Question regarding performance /bandwith using SSL tunnels..

    Posted 01-22-2008 05:19 AM
    Hi folks...

    This isn't an actual technical "problem" - more a general enquiry to see how other folks who use SSL tunnels are getting on.

    Back when I was running a pilot environment I just used static hubs and found that when doing Transfer Checks between hubs, I was easily able to hit pretty much the maximum possible bandwith for the switch. However now in our production environment, i'm using a medium strength tunnel (MEDIUM:!ADM) and on the same switches and infrastructure transfer rates have trickled right down to anywhere from 150KB/sec to 360KB/sec.

    I guess what I want to know is, just how much data or how many hosts do I need to have on that hub sending a constant stream of QoS information before the connection is saturated? With 31 servers chatting at the moment, my data_engine is recording anything up to 3 messages a second.. not exactly a cause for concern YET, but when should I be concerned?


  • 2.  Question regarding performance /bandwith using SSL tunnels..

    Posted 01-23-2008 05:27 AM
    We used to use a bunch of tunnels but being our env was pretty big we have switched to static hubs as this give better performance.  The ssl tunnels are a bigger hit on server perf more than bw I believe.  You may just want to check the performance of you hub as far as cpu and memory.

  • 3.  Question regarding performance /bandwith using SSL tunnels..

    Posted 01-23-2008 09:56 PM
    CPU and memory on the hubs is fine.. we deliberately way over specced them on the understanding that we might end up scaling to a monitoring environment about 5x larger than the scope of our initial pilot phase. We have one 'master' server which will be dishing out dashboards, etc and that is the tunnel server - the two hubs actually collecting all the monitoring data are tunnel clients. I've logged into each of the servers and they're barely registering any CPU load.. so we should be ok for now!

    Each server has space for a 2nd CPU but i'll be surprised if it any of them need more than they currently have (AMD Opteron 2214HE @ 2.2GHz).