We run a lot of EX3300 -48V DC switches at tower sites along with Ciena 3903 switches. Someone mentioned a concern about Juniper switches in uncontrolled environments. We keep them in a variety of cabinets but do not closely control temperature on them and have not had any issues with failures due to heat or cold. The majority of the cabinets are either open circulation (fan cooled) or heat exchangers. The cabinets with heat exchangers can reach 140F on hot days. The failures we have had with the Juniper switches have all been related to water damage. Yeah - stupid, but water manages to get into the damnedest places.
I’m not sure we would do the G.8032v2 with Ciena again. It’s worked well - but Ciena has a fair number of warts in the implementation that are probably more readily triggered in the WISP environment than in a typical fiber deployment. One particularly annoying one is when the licensed backhauls change modulation it occasionally corrupts packets - and since MEP packets are sent at a fairly high rate when it mangles one the Ciena at the far end sees a MEP packet that it doesn’t recognize and declares the link dead until manually reset. Not exactly helpful during widespread severe storms when the backhaul link topology is changing rapidly and repeatedly. Mark > On Jul 5, 2018, at 10:28 AM, Carl Peterson <[email protected]> wrote: > > Gino, We already use some MXs. Trying to avoid the cost and power > requirements to upgrade all our M7i routers to MX routers. Were you saying I > should look at doing a ring on MX or to keep it all routed on MX? If the > former, why not just do it on EX switches? The issue with MX for me is > primarily the routed 10G port cost. > > > Harold, Trill is interesting. What switches support it? Might look at it > for smaller networks off my core ring. > > > The management question is something that always needs to be considered, but > at least with Siklu, it is pretty easy to just run that out of band. We > generally do that anyway. Can't really test the alternate path without > falling over to it, but can monitor/graph the link performance. Thoughts on > all local management vs running a few management VLANs on the ring with a few > VRRP/HSRP management routers. > > > The real question for me is G.8032v2 vs ERPS (As Calix refers to it in their > docs) on Calix gear. I'm not sure if ERPS on Calix is G.8032v1 or some > proprietary version of that. Their docs show G.8032.v2 as being released for > multi-vendor compatibility which would indicate to me that their non v2 ERPS > config isn't multi-vendor compatible. I think I've settled on trying > G.8032v2 on Calix in the lab to shake out any issues. Being able to bring it > up with mostly 10G links and a 1G link is an easier sell and the ability to > add an MX into the ring could be quite useful down the road. > > > > On Thu, Jul 5, 2018 at 9:08 AM, Lewis Bergman <[email protected] > <mailto:[email protected]>> wrote: > Ok. That was pretty bad. It's the Lewis translator still active? > > > On Wed, Jul 4, 2018, 3:43 PM Rory Conaway <[email protected] > <mailto:[email protected]>> wrote: > Aren’t the Trill switches limited to 1Gbps so far? > > > > Rory > > > > From: AF [mailto:[email protected] <mailto:[email protected]>] On > Behalf Of Harold Bledsoe > Sent: Wednesday, July 4, 2018 12:59 PM > > > To: AnimalFarm Microwave Users Group > Subject: Re: [AFMUG] G.8032v2 or ERPS ring with 10G wireless links > > > > > Trill! 😀 > > > > On Tue, Jul 3, 2018 at 10:50 AM Carl Peterson <[email protected] > <mailto:[email protected]>> wrote: > > > > After moving most of our network over to VPLS, I'm working on pushing 10G > further into the network. It doesn't make a lot of sense to do this with > routers (cost & power) and since we use a lot of Calix gear for our GPON and > already have some of it in place it makes sense to me to just set up a 10G > ring and hang routers off of that where we need them. A couple of the legs > are going to be wireless 10G links for now so I'm looking for feedback on > G.8032v2 vs ERPS rings with wireless paths. > > > > Going to be running ~20-30 Service VLANs with multiple CVLANS and 20-30 other > single tagged VLANs over the ring (router to router, dedicated links, etc). > The only drawback I see is the inability to do traffic engineering the way we > an with MPLS VPLS but I'll still have that with the transport to the core > ring. > > > > Any gotchas I should look for? Issues with wireless? benefits of one vs the > other? > > > > > > > -- > AF mailing list > [email protected] <mailto:[email protected]> > http://af.afmug.com/mailman/listinfo/af_af.afmug.com > <http://af.afmug.com/mailman/listinfo/af_af.afmug.com> > -- > > Harold Bledsoe > > -- > AF mailing list > [email protected] <mailto:[email protected]> > http://af.afmug.com/mailman/listinfo/af_af.afmug.com > <http://af.afmug.com/mailman/listinfo/af_af.afmug.com> > > -- > AF mailing list > [email protected] <mailto:[email protected]> > http://af.afmug.com/mailman/listinfo/af_af.afmug.com > <http://af.afmug.com/mailman/listinfo/af_af.afmug.com> > > > > > -- > Carl Peterson > > PORT NETWORKS > > 401 E Pratt St, Ste 2553 > > Baltimore, MD 21202 > > (410) 637-3707 > > -- > AF mailing list > [email protected] > http://af.afmug.com/mailman/listinfo/af_af.afmug.com
-- AF mailing list [email protected] http://af.afmug.com/mailman/listinfo/af_af.afmug.com
