Wireless topology - WDS, CAPsMAN or something else?

Hi team!

After extending my network with two additional Mikrotik AP’s (AP4 and AP5 in the diagram) to fulfil necessary WiFi coverage I have met notable wireless traffic issues - transfer speed from WiFi clients in both 2.4 and 5 GHz bands to Public internet and local wired infrastructure has became 2-5 times lower and also sometimes there is a packet loss.

So I suppose the topology of network isn’t very good, probably there are loops. Any ideas how to solve the issue are welcome. Is it possible or even recommended to switch to CAPsMAN instead of WDS in this case?

In the diagram dotted lines with RJ45 are not existing yet, but they are only possible wired connections in this topology. Boxes mean buildings, lines marked as WDS mean WiFi visibility.
e9-mikrotik-wifi.drawio(1).png

Use cables everywhere you can.
It’s quite normal that the traffic e.g WDS AP1-AP4 eats the bandwith for AP1-AP2 so users of AP2 notice lower thorughput and so on.
Consider using OSPF to bypass loops and packets flowing different paths.

Do use AP1…AP2 to make interbuilding connections? Can you avoid 2.4/5 freqs and replace that missing interbuilding cables with
Wireless Wire https://mikrotik.com/product/wireless_wire to make these lines faster??

WDS is just a special case of “AP-bridge” - “station-bridge” implementation.

I removed WDS in an installed base once I discovered that

  1. WDS did not do packet aggregation (by design?) (MPDU aggregation is vital in effective use of wifi air-time. Airtime calculation with smaller AMPDU sizes explain why Mikrotiks legacy “wireless” has lower throughput and creates more “wait for transmit contention” lost air-time" than their “wifi” implementation.
    See links in : http://forum.mikrotik.com/t/wifiwave2-expected-performance/160075/1 for calculated loss in performance. No aggregation is dramatic for throughput.
  2. The old WDS installation, with multiple links, had selected one node as the WDS “master”. The WDS “master” keeps track of all client connections on the other nodes. However that selected WDS master node was on the weakest link, slowing down the whole WDS setup

So now I run “Wireless wire” for the backbone, and “AP-bride”-“station bridge” for the 2nd level connections. (branches)
That “AP bridge”-“station-bridge” connection uses NV2 protocol, (every branche is 1 AP and 3 stations) to slalom between the many IEEE802.11 transmitters (AP, station, clients)
The stations (SXTsq5ac) in the branches each ethernet-feed a local hAP and a wAP for client service.

Legacy Mikrotik has “AP-bridge”-“station-bridge”, the newer wifi has exactly the same functionality but that is named “AP” and “station-bridge”. (e.g. The bridge is always part of the AP mode)
Legacy WLAN interface and newer Wifi interface are not bridge compatible !

Footnote; CAPsMAN is more something on top of your topology: CAPWAP tunnels between the CAPsMAN controller and the CAP’s. Local delivery would require the same 4-address wifi links, central CAPsMAN traffic would allow to use normal 3-address wifi links between the AP’s. The same is true for any tunneled traffic in this topology.

I have used both WDS and CAPsMAN for wireless topologies, and each has its strengths. WDS is great for simple mesh networks, while CAPsMAN provides more centralized management and control over multiple access points. It really depends on the scale and complexity of your network setup.

Hi all and thanks for replies. Looks like my diagram hasn’t been posted, so I resized the image and am trying again.
e9-mikrotik-wifi.drawio-2.png

Hi! thanks for reply. but OSPF is a L3 protocol not L2, for now all issues are in layer 2

Hi and thanks! Using wires and disabling WDS on new AP’s solved the issue. But I’m still in doubt “what if” - if there wouldn’t be a possibility to use copper (or optics, or wireless wire) - what -should- be the correct architecture/topology for this setup.