-1

I have setup a wireless network at work office with the following config:

  • 192.168.1.1: TP-Link 941 Router, Root AP, connected to internet
  • 192.168.1.2: TP-Link 901 AP, WDS Repeater, connected to 192.168.1.1
  • 192.168.1.3: TP-Link 901 AP, WDS Repeater, connected to 192.168.1.1

All devices are WDS capable, working on channel 3, automatic version and encryption, 11bgn mixed. Until yesterday, everything was working fine. Ping times inside network (from my laptop to 192.168.1.1-3) were all less than 10ms even on heavy network loads. But from today morning, I see severe fluctuations in ping times, ranging from <1ms to complete timeouts. I am connected to 192.168.1.2 and even have problem pinging it. Because our internet link is 1Mbps in speed and we don't have any server(s) in our network, this can't be caused by heavy network loads. Any ideas about source of problem?

Mohammad Javahery
  • 9
  • 2
  • 7
  • Yep: WDS sucks. If you can, hard-wire directly to your APs. – EEAA Jan 27 '15 at 06:00
  • Thats the great thing about wireless. It allows anyone in your vacinity to ruin your bandwidth and latency! I'd be doing a site survey scan and looking at the cahnnels your APs are on, and moving them to less-congested channels if you can. – Mark Henderson Jan 27 '15 at 06:10
  • @EEAA I am planning to do it. You're right, WDS really sucks! – Mohammad Javahery Jan 27 '15 at 08:15
  • @MarkHenderson Changed channel to 7, got better results. But timeouts still appear and ping times are still high (some times up to 100ms). Can you introduce me a good and free tool for site survey (free because I am in Iran and cannot buy paid software)? – Mohammad Javahery Jan 27 '15 at 08:18
  • I'm using WDS widely, everything is perfect. What am I doing wrong ? Oh, guess, that's TP-Link that's sucking balls. You just have to stick to the right equipment, and throw away this junk. – drookie Jan 27 '15 at 11:22
  • @drookie I have used TP-Link equipments for years and nothing was wrong so far (Important: this is my first implementation of WDS totally). What equipment do you use? – Mohammad Javahery Jan 27 '15 at 11:42
  • Cisco/Juniper. The experience consists of what you have seen, not of what you haven't. – drookie Jan 28 '15 at 08:05
  • It's apparent that Cisco/Juniper equipments are better and give results which I have never seen. But when you are under severe budget constraints, you have no way other than sticking to brands such as TP-Link (or more awful as some of my friend stick to!) – Mohammad Javahery Jan 28 '15 at 09:35

2 Answers2

0

100ms is way too much. Something is clearly wrong. How is your security on the wireless part ?

What you could check/do:

  1. Try to add MAC filters and WPA (if not done so already) so your connections become as good as possible.

  2. Check the QoS settings in the router.

  3. Make sure there's no P2P heavy traffic (usually this is a problem for home routers due to too much open simultaneous connections). If you use P2P, try to limit the number of open connections to 80.

  4. Test the connection stability with blank connection (just ping -t something while having no traffic).

Overmind
  • 3,076
  • 2
  • 16
  • 25
  • Security is previously set to WPA/WPA2 Personal (automatic through negotiation between APs and clients). MAC filtering will be done as soon as possible (I planned it today). My APs are not QoS-capable, hence not applicable. No P2P or heavy traffic (as mentioned in orig. post, our internet link is only 1Mbps, so internet traffic can't be the cause). Numbers mentioned above are all given by ping, and vary time to time. – Mohammad Javahery Jan 27 '15 at 11:04
0

To investigate further, try to use tracert command to determine where exactly latency starts to increase a lot.

Run a tracert form a wirelessly-connected station and one from a station directly connected via UTP TP-Link 941. This will determine exactly where the latency problem comes form: either your wireless side or ISP or further ahead.

Also note that QoS can be configured in the main router even if APs don't support any. It acts like a filter/proxy, not like a spanning configuration. So if there is any, remove it from the configuration.

Overmind
  • 3,076
  • 2
  • 16
  • 25
  • I know that problem rises from my internal network, as these ping times are related to pinging from my wirelessly connected laptop to all APs inside my network (increasing as hop count to other APs increases, showing that problem exists in links between both repeaters and root AP, as well as between clients and APs). There is no QoS setting in 941 (root) router, only WMM setting exists and enabled. Bandwidth control is also disabled. – Mohammad Javahery Jan 27 '15 at 12:09
  • If APs allow it, try to disable the 40MHz channels to see if there's difference. Also, check the RTS/CTS if any. They can cause low speed/latency problems/ – Overmind Jan 27 '15 at 12:20
  • No luck with 40MHz channels. Still having timeouts and high latencies. What should be values of RTS/CTS? Currently set to defaults. – Mohammad Javahery Jan 28 '15 at 06:49
  • Recommended for the RTS threshold is around 500. A low threshold implies RTS packets are transferred more frequently and the throughput of the packet is on the lower side. But, sending more RTS packets can recover the network from collisions/interference/overload of traffic on a network. Best thing is to start form this value and test upper and lower values. A too lower value may introduce additional latency into the network; if you have many users far from the access point, try a value of 2304. After finding a stable value for RTS, next step is to work with Fragmentation. Default is 2346. – Overmind Jan 28 '15 at 07:38
  • Seems that default value (2346) for RTS is the best. Decreasing it (even to 2304) caused network to be completely unusable. Because APs and router are all 11n-capable, fragmentation threshold cannot be modified. Any more suggestion(s)? – Mohammad Javahery Jan 28 '15 at 09:28