wAP AC 3 (IEEE 802.1ax)

Do you want wAP AC 3 in 2019?

  • Yes!
  • No!
  • Forget wAP AC 3, I want ROS7!
  • I want a burger
0 voters

Hello,

we have hAP AC2 and we have cAP AC with 802.11ac Wave 2 (2x2:2) IPQ4018/IPQ4019 chipsets. Well what´s next? IEEE 802.1ax of course
Reading https://wikidevi.com/wiki/Qualcomm_Atheros we have a nice ASIC:

IPQ8072
Wave 3, MU-MIMO, 1024 QAM,
OFDMA, 4x GbE, 2x 10G (NBASE-T)
Advanced QoS, TxBF, Wi-Fi SON,
14nm (6.5 Gbps max. PHY rate)
So, 4x4:4 with 2.4Ghz and 4x4:4 with 5.0Ghz with wAP AC design

Therefore I would like to suggest a new wAP AC 3 accesspoint for 2019 with mentioned ASIC. What do you think?

the bigger problem is driver support since Mikrotik creates here own drivers.
The actual drivers doesn’t support anything of WAVE 2, are way behind competitors Performance,
and this will not change, so I’m not Interested in new devices, with rudimentary driver support and without any features…

v7 is only a number. What do you expect from a number change?
All I expect from MIKROTIK is the NV2 working and soho ap. Version numbers are not interested.

The number is regardless
The problem is behind the numbers. ROS 6 is based on Linux Kernel 3.XX.XX
And all new drivers from the Manufactures are only running and supporting Kernel 4.xx
Does Not pushing ROS 7 will end in disaster.

Look at the problems with IPQ4018

100% agree. I have no idea why Mikrotik insists on making their own drivers. We get poor performance, missing basic features (802.11w, Wave 2 support, etc) and for what benefit? By the time a new AP / chipset comes out with decent drivers the technology has already moved on.

I think I will be looking at different vendors when it’s time to deploy 802.1ax, it will be years before a Mikrotik solution is ready.

which competitors?

same cost??

How long list you want to get?
What matters cost if product performs half throughput then overs with identical Wireless Chipset??
what’s the savings when you have trouble with the customer?

which vendors??

same cost?

cost is always the most predominant and determinant item on all project

There is a lot that is expected from v7 beyond the fluffy stuff Mikotik keep back porting to v6.

Just a small selection:

4.x Linux Kernel

  • IPv6 improvements
  • Removal of “route cache” that causes so many issues
  • VXLAN support
  • Ease of integration with SDK’s, e.g. OpenFastPath/OpenDataPlane, DPDK, Marvell SDK’s
  • Wireguard VPN support

New routing engine

  • Improved route filter processing performance
  • Improved route search/view performance
  • Lower memory footprint for storage of routes
  • Ability to select per BGP instance if routes are installed in the FIB, or are RIB only. Allowing RouterOS to act as a true Route Reflector
  • As above but as a filter option, “Do Not Install in FIB”
  • VPNv4 NLRI Bug Fix
  • Support for VPNv6
  • “Add to Address-List” route filter option
  • AS range matching support in route filters
  • Extended Dynamic, Static, DHCP, PPP route origin matching in route filters
  • OSPFv3 fixes for Loopback addresses
  • True VRF support with complete isolation between VRF’s and ability to terminate “services” inside a VRF, e.g. BGP, OSPF, L2TP, WinBox, SSH.
  • MPLS Fast Re-Route Support
  • eVPN support


    I honestly think if releasing v7 meant dropping support for the Tilera based CCR’s in v7, it would be worth it.. They are 5 years old after all. Some customers would be a bit annoyed at first, but would soon get over it.

“Once you’re done changing, you’re done” - Jack Tramiel

Pushing on later certainly will not help

MkroTik doesn’t create its own drivers. They get a driver package/source from vendors and Integrate it into RouterOS. The question is whether those driver need need some kind of features/connectors the Linux kernel of RouterOS doesn’t provide and therefore MikroTik has to modify them to work with RouterOS without any problems.

False, they use there own driver, based on open source and not the closed source one from Qualcomm, that is the reason we missing Spectral Scan etc…

Here is your list:

UBNT
Ligowave
Zyxel
Cisco
Cambium
Ruckus
Ignite
Lancom
ZTE
HP Wireless

These are all Controller or Hotspot Systems we use, Mikrotik has really great devices but they don’t have Good Wireless Trougput !!

rb951G 2.4ghz vs laptop w8.1pro intel n7260

wifi-speed.png

@chechito go on with 5 Ghz and do the same with 20 simultaneously clients

This is where you are incorrect

I’m no authority on the matter, but my understanding is that Mikrotik use a closed source driver that uses code from Qualcomm as well as their own proprietary “special sauce” to delivery custom Media Access Control schemes like Nstream and NV2.

The reason for the lack of spectral scan is that in the 802.11ac generation of chipsets this function was not universal accross all Qualcomm Atheros chips, on top of that I believe that ABI changed between the older driver used for 802.11n and the new 802.11ac driver.

Best to avoid anything open source and re-invent everything in house?

Meanwhile every other manufacturer is happily using ath10k driver:
https://wireless.wiki.kernel.org/en/users/drivers/ath10k

And yes, open source driver even has working spectral scan…
https://wireless.wiki.kernel.org/en/users/drivers/ath10k/spectral

I really want to love Mikrotik products due to the ease of configuration and deployment combined with the great hardware and value but sometimes these software decisions feel like bashing my head into a brick wall…

currently i dont have access to equipment in that conditions, but i can say with the proper design and proper config you can operate high density scenarios with up to 40 client devices per radio with good performance, this scenarios require good troubleshooting and fine tuning of all the flexible options mikrotik offers, proper configured and designed underlying infrastructure and of course make a good QoS implementation, with many client devices all this conditions are mandatory to get successful wifi