The MPLS WG Archive[Date Prev][Date Next][Thread Prev][Thread Next] [Date Index][Thread Index][Author Index][Subject Index] draft-meyer-mpls-soft-preemption-00.txt
In message <20030221030857.GJ18082@gblx.net>, Matthew Meyer writes: > Comments in line. > > Thus spake Markus Jork (mjork@avici.com): > > |I like the original soft preemption idea in the draft as it stands now. > > I think there have been some strong arguments to leave it as is. > > |But I have my doubts about the extensions discussed in this e-mail thread. > |When planning an admin link or node shutdown, it is probably a better > |idea to indicate that via the IGP (through a metric change or announcing > |the available bandwidth as 0). > > While I agree with you that it is perhaps more appropriately > accomplished in the IGP, I don't believe the IGP has all > the needed mechanisms. > > Cranking the metric & setting/flooding available reserveable BW > at 0 is not enough. Even if you soft preempt as well, you still > have 0BW LSPs or explicitly routed LSPs or disparate backups that > need to be forced to forget this link. Currently SPs are forced > to remove configuration parameters if we want to drive some types > of tunnels off a link. Keeping a admin-color (aka admin-group) and naming it "maintenance" and always configuring exclude-admin-color maintenance to your LSPs does solve this problem easily. Just advertise the link with "admin-color maintenance" and the LSPs should go away rather quickly. This is just a good operations trick/practice to have handy. Increasing the IGP cost keeps the non-MPLS packets away. Curtis
|
|