Let’s say we have two machine cluster, each with 2x10C.
With bare-metal we get up to 500 CCPS (no radius but with a simple ruby script).
We saturate only 5 out of 20 cores under 500 CCPS (3 for apps + 2 for DPDK)
Gateway app hits 100% quota under this load.
Is there a way to optimize the config to go beyond that other than doing multiple clusters under KVM on this 2 server setup ?
I know that getting a next gen CPU can get twice the number but I would like to know if there are GENERAL ways I can tweak it.
In those conditions, gateway application must not hit 100%. Would you contact with the support team from support@telcobridges.com, please? Please include tbreport into your email.
Hi,
some information after more testing and correcting Pawel a bit:
we used dell R620 with 2x E5-2643V2 so 2x6c not 2x10c.
After correcting routing (in first attempt we always tried first route with unavailable NAP…) and tweaking testing scenario we got about 750 CCPS with engine and gateway applications both reaching 100%. Total cpu usage was about 65%.
We only tested sip without rtp streams.
So comparing this with results on wiki we think it quite good.
