Hi Joe, On Wed, Aug 28, 2019 at 8:53 PM Joe Clarke (jclarke) <jclarke@xxxxxxxxx> wrote: > >> === > >> > >> In section 6.6, do you have any more concrete recommendations on a reasonable > >> limit of LSPs with auto-bandwidth that you have discovered from testing or > >> operational experience? Providing some data here may prove useful, even if it > >> is somewhat anecdotal. > >> > >> > > > > We discussed this and felt reluctance in putting a number down in the > > RFC, esp when we don't do it for our base published RFCs (ex. such as > > how many LSPs can be delegated to a PCE). I also checked with at least > > one vendor and was told that it is quite dependent on the deployment > > scenario and would let the operator decide. > > Maybe not a hard number, but is there any kind of example you could provide for a given scenario that provides clearer guidance on what kind of impact each one of these LSPs would present? > How is this - 6.6. Impact On Network Operations In order to avoid any unacceptable impact on network operations, an implementation SHOULD allow a limit to be placed on the number of LSPs that can be enabled with auto-bandwidth feature. For each LSP enabled with auto-bandwidth feature there is an extra load on PCC, as it needs to monitor the traffic and report the calculated bandwidth to be adjusted to the PCE. The PCE further re-compute paths based on the requested bandwidth and update the path to the PCC, which in turns triggers the re-signalling of the path. All these steps adds extra load and churn in the network and thus operator needs to take due care while enabling this features on a number of LSPs. Thanks! Dhruv > Joe >
