Firmware Road Map

[drbaker]

Any news about MSTP?

[sirhc]

MSTP is next after we get all the current features working 100%

After MSTP we are nearing the end I think of Large feature sets?

[tma]

> After MSTP we are nearing the end I think of Large feature sets?

Well, I don't know whether your chipset can do QoS in terms of traffic shaping - like prioritizing and throttling based on port and IP address ranges. This is something Linux is really good at, and until recently, I wouldn't have thought that traffic shaping should be done on switches. However, given the unfortunate fact that the Cavium-based EdgeMax routers will do traffic shaping only when IP offload is turned off (which makes them really slow), I'm in a need for a device in the backbone (on each station) that can help out - that would be the switch(es).

So if your question was meant as an invitation to give you another large feature set to work on, here you go ...

[sirhc]

> After MSTP we are nearing the end I think of Large feature sets?

Well, I don't know whether your chipset can do QoS in terms of traffic shaping - like prioritizing and throttling based on port and IP address ranges. This is something Linux is really good at, and until recently, I wouldn't have thought that traffic shaping should be done on switches. However, given the unfortunate fact that the Cavium-based EdgeMax routers will do traffic shaping only when IP offload is turned off (which makes them really slow), I'm in a need for a device in the backbone (on each station) that can help out - that would be the switch(es).

So if your question was meant as an invitation to give you another large feature set to work on, here you go ...

Our next family of switch can do this. NOT COMING OUT ANY TIME SOON THOUGH

[tonym]

As long as said layer 3 device is in all VLANs. There is a blurry line where modern switches are really layer 2.5

Take a look at all features and ask yourself what layer they are. It can be surprising.

DHCP relay is still a handy feature...

...once again a balancing act.

I have a feature similar to DHCP relay on some of my layer 2 devices. It is called MAC Forced Forwarding. It takes broadcasts and directs them at the l3 gateway as detected by either DHCP snooping or explicit definition. It also uses a form of ARP proxy to reply to all requests with the address of the router even if the device is local. This reduces the amount of broadcasts traversing the network and contains issues. The only problem is i'm not sure how it interacts with APs connected to a switch doing MACFF, maybe it would only work properly if client isolation is enabled.

I would want MACFF over DHCP relay personally.

https://en.wikipedia.org/wiki/MAC-Forced_Forwarding

[Eric Stern]

As long as said layer 3 device is in all VLANs. There is a blurry line where modern switches are really layer 2.5

Take a look at all features and ask yourself what layer they are. It can be surprising.

DHCP relay is still a handy feature...

...once again a balancing act.

I have a feature similar to DHCP relay on some of my layer 2 devices. It is called MAC Forced Forwarding. It takes broadcasts and directs them at the l3 gateway as detected by either DHCP snooping or explicit definition. It also uses a form of ARP proxy to reply to all requests with the address of the router even if the device is local. This reduces the amount of broadcasts traversing the network and contains issues. The only problem is i'm not sure how it interacts with APs connected to a switch doing MACFF, maybe it would only work properly if client isolation is enabled.

I would want MACFF over DHCP relay personally.

https://en.wikipedia.org/wiki/MAC-Forced_Forwarding

I've added this to our future "to-do" list.

[Travis_WVN]

Any chance for MPLS-VPLS type features, or even better going up to TRILL or SPB (802.1aq)? I know those are usually data-center feature-sets, but it has applicability for the WISP industry if you consider the performance/reliability benefits of having a large number of cross-connects between tower sites in dense areas.

[mike99]

Any chance for MPLS-VPLS type features

Q in Q but I think Eric would have to look to be able to also encapsulated protocols.

[Eric Stern]

Any chance for MPLS-VPLS type features, or even better going up to TRILL or SPB (802.1aq)? I know those are usually data-center feature-sets, but it has applicability for the WISP industry if you consider the performance/reliability benefits of having a large number of cross-connects between tower sites in dense areas.

This won't be possible as our switch core does not support MPLS.

[sirhc]

However our next generation switch cores do support MPLS, SFP+, and routing but there is no schedule on these models so please do not ask when as the answer would be I DO NOT KNOW AN ETA