Hi Ben, My experience with MX (and other vendors) ERPS is that you will get very good convergence times without enabling OAM at all. Unless you have intermediate non-ERPS capable switches that require the OAM or intermediate xWDM transponders without fault reflection etc then I would recommend testing it without OAM configured and see what your results are.
Cheers, Caillin -----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Ben Boyd Sent: Tuesday, 5 June 2012 5:03 AM To: Ben Dale Cc: [email protected] Subject: Re: [j-nsp] EX/MX G.8032 with OAM CCM's Here is config from one of the MX nodes: oam { ethernet { connectivity-fault-management { action-profile rmep-defaults { default-actions { interface-down; } } maintenance-domain d1 { level 0; maintenance-association control { inactive: continuity-check { interval 10ms; loss-threshold 5; hold-interval 1; } mep 2 { interface xe-2/0/0; remote-mep 1 { action-profile rmep-defaults; } } } } maintenance-domain d8 { level 0; maintenance-association control { inactive: continuity-check { interval 10ms; loss-threshold 5; hold-interval 1; } mep 1 { interface xe-1/3/0.16; remote-mep 2 { action-profile rmep-defaults; } } } } } } } protection-group { ethernet-ring sidera-ring-1 { east-interface { control-channel { xe-1/3/0.10; } } west-interface { control-channel { xe-2/0/0.10; } } data-channel { vlan 1-4094; } } } Here is one of the configs from the EX nodes: protection-group { ethernet-ring sidera-ring-1 { east-interface { control-channel { xe-0/1/0.0; } } west-interface { control-channel { xe-0/1/2.0; } } control-vlan MANAGEMENT; data-channel { vlan [ MANAGEMENT CUSTOMER_A CUSTOMER_B CUSTOMER_C CUSTOMER_D_TRUNK L2TUNNEL ]; } } } oam { ethernet { connectivity-fault-management { action-profile rmep-defaults { default-actions { interface-down; } } maintenance-domain d5 { level 0; maintenance-association control { inactive: continuity-check { interval 10ms; loss-threshold 5; hold-interval 1; } mep 2 { interface xe-0/1/0.0 vlan-id 13; remote-mep 1 { action-profile rmep-defaults; } } } } maintenance-domain d6 { level 0; maintenance-association control { inactive: continuity-check { interval 10ms; loss-threshold 5; hold-interval 1; } mep 1 { interface xe-0/1/2.0 vlan-id 14; remote-mep 2 { action-profile rmep-defaults; } } } } } } } When OAM is activated, several of my interfaces in the ring start bouncing...... I'm wondering if the EX's just can't handle the 10ms interval. Customer is looking for sub 50ms convergence times (competing with REP on Ciscos). We skipped configuring it, but if we make it back, I'll set it up to 400ms x 3 and see if that changes anything..... but wouldn't RSTP just as fast in this case? ----------------------------------- Ben Boyd [email protected] http://about.me/benboyd On Jun 2, 2012, at 6:27 PM, Ben Dale wrote: > Having just done this in the last week (strictly EXs), I'd suggest that neither your OAM and G.8032 are configured correctly - post up some configuration if you can. > > Make sure you have your link-fault-management action-profile is set to link-down and that you have an apply-action-profile bound to the interface as well. > > My testing (very rough) showed about 0.5 seconds between link down and switch-over - I chalk this up to my test lab being copper interfaces and the slight lag in link-down detection. > > I think the minimum timers for OAM on the EX are 400ms x 3, which is nothing special, so unless you have intermediate Layer 1 devices to detect failure in, I don't think mixing OAM and G.8032 really buys you BFD-like performance. > > I'm hoping this will be improved in the future - I seem to recall Extreme's EAPS clocking in at a order of magnitude or two quicker. > > Cheers, > > Ben > > On 01/06/2012, at 10:21 PM, Ben Boyd wrote: > >> Experts, >> >> Does anyone out there have any working G.8032 ethernet ring w/ OAM backend configurations/implementations? >> >> >> I setup an 8 switch (EX/MX mix) ring in the lab and OAM seems to be complicating things rather than acting like a layer 2 BFD. >> >> >> When I fail a node (not the ring owner) without OAM configured in >> just a plain G.8032 ring, the convergence time is around 30 seconds. >> (Seems a bit high. I've seen better times with RSTP) >> >> When I fail a node (not the ring owner) with OAM configured, 50-70% of the ring goes to discarding and fluctuates from there, never ultimately converging. >> >> Link failures on the other hand, converge in about 45ms. >> >> >> >> If you need configs I can post them, but they're pretty vanilla (from the config examples on Juniper's site). >> >> >> >> ----------------------------------- >> Ben Boyd >> [email protected] >> http://about.me/benboyd >> >> >> >> >> _______________________________________________ >> juniper-nsp mailing list [email protected] >> https://puck.nether.net/mailman/listinfo/juniper-nsp >> > _______________________________________________ juniper-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/juniper-nsp -- Message protected by MailGuard: e-mail anti-virus, anti-spam and content filtering.http://www.mailguard.com.au/mg _______________________________________________ juniper-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/juniper-nsp
