Hello there,
does anybody know, if Mikrotik is planning to produce 60ghz radios with 802.11ay chipset and support for full frequency band, eventually with support for half-channel?
The only ones to know are Mikrotik staff and they don't make it a habit to reveal their HW roadmap here on the forum.
thats pitty, they arent annoucing future plans. i was hoping that after all the year they change chipset which unlocks all potencial of 60g technology. for sure, newer chipsets from Peraso or Qualcomm are more expensive. FW development, testing will take time and more expenses, but they can keep up with competition. thats just my opinion. hope somebody from staff will read this topic.
I would be VERY interested. This will be a bit of an article but I want to give a full idea of what’s going one.
I’m using Wave and Tachyon now. Wave for longer links in low density, and Tachyon for high density. I’m not trying to promote these, just saying I’m heavy into mmWave tech.
So here’s a current model with Tachyon. This is a terragraph work-alike made possible by Mikrotik, especially newer routeros versions.
I cluster 4 Tachyon APs for 360 coverage. These get an ABAB channel plan and an SSID per that matches the nearest of 8 cardinal directions, ie n/s/e/w/ne/se/nw/ne. Tachyon CPE can have 8 SSIDs assigned so they can all be pre-setup. The site gets 1-3 tachyon radios for backhaul. These have 2.5G ports and >2Gbps HDX capacity.
I then use a variety of 2.5G switch options (I’m desperate for a new outdoor 2.5G switch….) and an rb5009.
The APs are in a bridge and the other ports are not. OSPFv3 with LL addressing and some sensible defaults, so each individual unbridged port is seen by OSPFv3 as distinct and the bridge has a single LL IP. Each site gets a dedicated port for backhaul to 1-3 other sites via OSPFv3 through the bridge/APs at a parent or peer site.
This makes for an almost self-configuring and self-healing network. Plus I can add other links in and just set a much higher cost so there’s a non-mmwave last hope route.
since mikrotik introduced ipv6 nexthops on ipv4, now I don’t even run a separate routing over this OR do the GRE tunnels I was doing. I just let ipv6 do it all and it’s really good at it. rb5009 can easily handle the throughput and even run fq-codel on interfaces. full dual stack. Very good, very stable, fast, link drop tollerant etc.
So here’s the thing, if I could see a Mikrotik Peraso model or similar ‘good’ ad/ay radio but had some features like a chip similar or a bit faster than the rb5009's, and at least 3 2.5-10G ethernet ports that can do both PoE in and out and enough power at 48V to run a ‘ring’ of these radios off a ‘bt power supply, I’d be in heaven. ideally a wifi6/7 radio as well.
I could replicate the current mikrotik+tachyon design WITHOUT having extra switches etc, and driven by one or more cables (PoE-in on just 2 devices creates a redundant DC ring also). CLEAN.
ospfv3 onboard. cake and fq_codel onboard. dhcp-onboard. ideally a wifi6/7 radio as well. really just does everything. note that vxlan ipv4 works perfectly over the ipv6 next-hops as well and if you layered on iBGP just for EVPN you’d get that too.
one AP SKU, one CPE sku which is more like the current ‘ay cube CPE form factor. Instead of more SKUs, make some 3x and 4x mounting kits.
slam dunk. please just stay away from qualcomm. peraso would be great right now.
2 Likes
Agreed. A MikroTik + Peraso unit would be a match made in heaven. Peraso modules interface with their hosts via USB-C (and custom power levels), so integration with MikroTik hardware could be modular, on an as-needed basis.
An RB5009 brain with a couple ethernet ports and 1-3 Peraso modules would be killer.
It’s an interesting idea to just have modular external beamforming radio-antennas. Just need a separate USB3 controller for each radio. Peraso gives you the SDK access, even a lowly tinkerer like me can get them to build the drivers.
rb5009 with 6-8 USB3.0 controllers and Peraso drivers would be hillarious.
Unfortunately it looksl ike the USB ports on basically everything outdoor today is a USB2.0 port per the block diagrams. (netmetal ax, mantbox ax, etc)
BUT, the rb5009’s 88F7040 chip has a pcie 3.0x4 controller, 2 3.0x1, 2x USB3.0. Which means mikrotik already has a mature platform they use today that could do this, it just needs a USB3.0 controller added to support the radios and of course some packaging and software dev for driver integrations.
I really love this idea though. A device at a similar or somewhat better performance level to rb5009 and a row of 6-8 USB3.0 ports and then 2 choices of external radio, a wide 90-120 scanning beam, and a 20-40 scanning beam. Those could be put right in the the sxtsq body which would be super flexible.
Anyone at mikrotik listening? this could actually be really amazing.
2 Likes