1510 MTU on EP-R6 and ER-X ubiquiti, stay away!

[rebelwireless]

yeah, so I got an EP-R6 to play with. I like the concept, very slick, yada yada. Until I try to set the MTU! maximum of 1510 supported! WTF. This is a brand-new device, sold in 2016, and it doesn't handle MTU above 1510.

That's not enough for ANY tunnel like GRE, EoIP, IPIP, PPTP, L2TP.

Yeah, you can slide MPLS/VPLS over that in 8bytes, but they you can't run a VLAN across the VPLS.

Note that this is NOT in the spec sheet of the device.

somebody slap me the next time I but anything but a radio from ubiquiti.

[rebelwireless]

minor correction, 8beta3 firmware has this up to 2000. who knows when that will be released though. doesn't change that I have EP-R6 that are useless to me right now. yet another ubiquiti device collecting dust.

[mike99]

Ubnt is full of good idea but in the end, not really what we need. If only Netonix could do a outdoor layer 3 switch with multiple internal NIC with flow control to tag, or untag, VLANs to. This is what edgepoint should have been, that would be the kind THE product we need.

Even for radio, I'm looking a lot at Cambium even if new Ubnt AC radio with GPS sync could be nice (if it's not like TI GPS sync :-/ ). Still, Cambium will have beam steering soon (ePMP 2000). In my opinion, the only great product Ubnt have is the AF line. Maybe a PtMP version of AFX could change my mind about Ubnt radios.

[sirhc]

Lets just say that a Netonix "Router" and Layer 3/4 switches "could" already be in the works?!

If we were to make a router and I am not saying we are it would be based on VYOS.

If we were and I am not saying we are making a Layer 3/4 Switch it would continue to use the VITESSE switch cores which I think we can all say are excellent quality.

Our Layer 3/4 switch (if we were making one) would be capable of OSPF with BFD but not BGP world routes (too many routes for core memory).

[rebelwireless]

an OSPF capable WS-12-250-DC would be a highly desirable thing....

VYOS implies no MPLS though... :(

Unless of course you guys wanted to invent your own rflow and some type of tunnel in the switch fabric....

[rebelwireless]

actually, just supporting a GRE or EoIP type tunnel in the device with an easy way to tie that to a port or vlan would do it. I could very much live with a 100% layer3/4 wispswitch w/ OSPF+BFD and a layer2 tunnel protocol on-device. (MTU large enough to handle 1508+tunnel overhead and support QinQ)

[sirhc]

actually, just supporting a GRE or EoIP type tunnel in the device with an easy way to tie that to a port or vlan would do it. I could very much live with a 100% layer3/4 wispswitch w/ OSPF+BFD and a layer2 tunnel protocol on-device. (MTU large enough to handle 1508+tunnel overhead and support QinQ)

Actually the Layer 3/4 switch core from Vitesse is MPLS capable if we opt for the more expensive core for production? The different cores all use the same package meaning the chips look the same on the outside and more than likely are the same on the inside but unless you pay for the features they are shut off or so they say.

The more expensive core is actually capable of a LOT more stuff.

Same thing when it comes to the switch core we use on our current line, they sale you different models such as VSC-7424, 7425, and 7427. They all look the same except the printing on the chip for the model number, "supposedly" the smaller numbers are not supposed to have as many ports....but in reality they do! But we sign an agreement that we do not cheat so we do not.

We use:
VSC-7424 for 5-10 ports (do not ask me about the 5 port - shhhh)
VSC-7425 for 12-14 ports
VSC-7427 for 24 ports

[rebelwireless]

well.....that would be something interesting to consider. I think the WS-12-x models are pretty awesome. If you could simply license up those models to be layer3/4+MPLS capable without any physical re-engineering, I think they would be EXCEPTIONAL.

I would have a single Netonix, running OSPF+BFD between backhauls, with LDP/MPLS, and be extremely happy!

I should say that I'm implying VPLS and TE here. Really need to have both VPLS and TE for MPLS to be of any use.

[sirhc]

well.....that would be something interesting to consider. I think the WS-12-x models are pretty awesome. If you could simply license up those models to be layer3/4+MPLS capable without any physical re-engineering, I think they would be EXCEPTIONAL.

I would have a single Netonix, running OSPF+BFD between backhauls, with LDP/MPLS, and be extremely happy!

I should say that I'm implying VPLS and TE here. Really need to have both VPLS and TE for MPLS to be of any use.

No, you mis-understand.

The current models are based on the VSC-742X family switch core / SOC

Newer product line is a whole other family of switch core / SOC

We can not simply change the existing models with licensing but on the newer family of cores we are developing there is differnt levels in the same package meaning once we design the board we can simply drop in a different package (SOC) for more features.

[rebelwireless]

understood. maybe better to say that because the WS-12-x models are such a great fit for most of us, please make a VERY similar model with the added layer3/4 stuff :)