Hi Folks I've added the suggested info to the following wiki https://wiki.opnfv.org/display/vsperf/VSPERF+Deployment+Topology+Description+Scheme for review later today
Thanks Maryam From: opnfv-tech-discuss-boun...@lists.opnfv.org [mailto:opnfv-tech-discuss-boun...@lists.opnfv.org] On Behalf Of Tahhan, Maryam Sent: Tuesday, September 6, 2016 5:15 PM To: Klozik, MartinX <martinx.klo...@intel.com>; opnfv-tech-discuss@lists.opnfv.org Subject: Re: [opnfv-tech-discuss] [VSPERF] topology description scheme Hi Martin Comment inline From: Klozik, MartinX Sent: Tuesday, September 6, 2016 3:16 PM To: Tahhan, Maryam <maryam.tah...@intel.com<mailto:maryam.tah...@intel.com>>; opnfv-tech-discuss@lists.opnfv.org<mailto:opnfv-tech-discuss@lists.opnfv.org> Subject: RE: [VSPERF] topology description scheme Hi Maryam, Thanks for initial proposal. It looks good overall. It seems, we are going to define a grammar of a small language. In that case it might be useful to define it well for both human and machine processing. Just for case, we would like to create a class, which will be able to prepare a setup automatically based on the scheme name. In that case, we should consider: 1) Possible ambiguity in number meaning; May be, we should consider a standalone number as an index and number prefixed with x as multiplication factor, e.g. LS1 vs. PVx3P <MT> sounds good 2) In your proposal is V a synonym for VM with 2 NICs; We should consider a possibility of VMs with different NIC number, i.e. to really use V as a Virtual NIC, e.g. PVVP in meaning of phy_nic->virt_nic->virt_nic->phy_nic. However I can't figure out an intuitive but still flexible way to describe complex scenarios, e.g. 1VM with 4NICs in parallel or even 2 VMs in series, where 1st VM has 2NICs, but 2nd 4NICs in parallel... <MT> Just to be clear the proposal suggests V is a virtual NIC not a VNF... so the example for PVP and PVVP was incorrect below... I meant to update that before sending out the mail and I forgot... :$ PVP LS_PV2P PVVP LS_PV4P Best Regards, Martin From: opnfv-tech-discuss-boun...@lists.opnfv.org<mailto:opnfv-tech-discuss-boun...@lists.opnfv.org> [mailto:opnfv-tech-discuss-boun...@lists.opnfv.org] On Behalf Of Tahhan, Maryam Sent: Tuesday, September 06, 2016 11:21 AM To: opnfv-tech-discuss@lists.opnfv.org<mailto:opnfv-tech-discuss@lists.opnfv.org> Subject: [opnfv-tech-discuss] [VSPERF] topology description scheme Hi folks I've been thinking about this and I think I have a possible suggestion Lettering: It's case sensitive L - Local R - Remote S - vSwitch P- Physical NIC V - Virtual NIC pp - patch port O - Overlay port - not sure if we need... Numbering: Numeric values can follow any of the last S, V, and P or nothing follows them which implies a 1. For ports it implies the number of ports V2 is two virtual ports connected in succession. For switches it actually acts as an Identifier LS2 is local switch 2. The scheme is really to describe the flows from port to port through the vswitch so Current scheme Proposed scheme Phy2phy LS_P2 PVP LS_PVP PVVP LS_PV2P Multiple VMs in series LS_PVxP where X is the number Multiple VMs in Parallel LS_PVP_x (where X is a number applies to everything surrounded by underscores) 2 local switches doing phy2phy LS1_P2_LS2_P2 2 local switches patched together LS1_Ppp_LS2_P Overlayed -switches on a local and a remote LS_PO_RS_OP Overlayed - local switched LS1_PO_LS2_OP Not as simple as I would have hoped, but open to any suggestions BR Marayam
_______________________________________________ opnfv-tech-discuss mailing list opnfv-tech-discuss@lists.opnfv.org https://lists.opnfv.org/mailman/listinfo/opnfv-tech-discuss