AiMesh Newbie - will it improve latency vs an Access Point

[gbguy71]

I currently have an AC68U (running Merlin for OpenVPN) connected to a Deco M5 mesh network for wifi coverage. The Decos provide needed coverage for the entire house.

The Deco network is configured as an access point. I have one streaming service that is sensitive to the extra latency the Deco access point introduces (pings take roughly double vs. connecting to the AC68U's wifi)

If I get rid of the Decos and build an AiMesh network will I get similar latency as a single router or will it be similar to an access point?

 

[Livin]

I think you are asking... if you use the exact same 'star network architecture' you are today but replace the Deco APs with ASUS -- Will latency reduce.
Definitely no unless ASUS is now using quantum entanglement. Or unless the Deco's have latency problems other routers do not.

More Hops = More Latency; cant change physics.

If you can improve the connection speeds to gain real-world improvements (like going from wireless to wired backhaul) then you will see a latency reduction. Going from AC to AX backhaul might show some improvement, but as with everything, YMMV

 

[gbguy71]

Thanks. I didn't make myself clear. I don't understand how AiMesh works. I don't know if it has any latency advantages over my current setup.

Right now I have:
Cable modem <---> AC68U <----> (four Deco M5s). The M5s provide the access point that I use to access wifi.

I'm curious if:
Cable modem <---> (three AiMesh routers) with the AiMesh collection providing wifi access provides any latency advantages.

I guess the basic question is, "Does it reduce the number of hops?"

 

[Tech Junky]

SMH... Mesh is dumb and doesn't work well.

CM <> Wired router / AP's for WIFI

If you don't have Ethernet wired to the locations where you're wanting faster WIFI get it wired and you'll have less headaches.

Latency to the internal network should maybe be 1ms or less. A ping from the router say 8.8.8.8 (google) should be ~20ms or so. It depends on some other things which would show up in a traceroute to wherever you're trying to get to.

server:~$ traceroute 8.8.8.8
traceroute to 8.8.8.8 (8.8.8.8), 30 hops max, 60 byte packets
 1  192.168.12.1 (192.168.12.1)  0.801 ms  0.726 ms  0.697 ms
 2  192.0.0.1 (192.0.0.1)  1.005 ms  0.984 ms  0.963 ms  (TMHI)
 3  * * *
 4  * * *
 5  * * *
 6  192.0.0.1 (192.0.0.1)  21.755 ms * *
 7  192.0.0.1 (192.0.0.1)  19.248 ms  17.843 ms 10.160.174.201 (10.160.174.201)  18.480 ms
 8  10.164.176.13 (10.164.176.13)  23.528 ms  49.728 ms 10.160.174.201 (10.160.174.201)  45.185 ms (exiting TMHI)
 9  72.14.242.140 (72.14.242.140)  61.723 ms  61.103 ms 10.164.176.13 (10.164.176.13)  43.931 ms
10  72.14.242.140 (72.14.242.140)  31.402 ms 108.170.252.129 (108.170.252.129)  30.409 ms  30.337 ms
11  142.251.60.139 (142.251.60.139)  28.366 ms 142.251.60.45 (142.251.60.45)  39.812 ms 108.170.252.161 (108.170.252.161)  34.749 ms
12  72.14.237.47 (72.14.237.47)  35.992 ms 209.85.243.255 (209.85.243.255)  37.737 ms dns.google (8.8.8.8)  26.000 ms

So, you see I have to contend with ~7 hops just to get out of TMHI to the real world. Now, on cable IIRC it was still maybe 4-5 hops before hitting the real world. The path in which the provider takes adds latency just as much as the speed of the connection. The first hop is under 1ms to hit the internal ISP network. Since there's a conversion from IPv4 to v6 after that it's masked but you can see it's added ~20-40ms before it exits to the real world.

From Windows it's an extra couple of hops.

Tracing route to dns.google [8.8.8.8]
over a maximum of 30 hops:

  1     1 ms     1 ms     1 ms  SERVER [192.168.0.1]
  2     2 ms     2 ms     2 ms  192.168.12.1
  3     2 ms     2 ms     2 ms  192.0.0.1
  4     *        *        *     Request timed out.
  5    18 ms     *        *     192.0.0.1
  6    34 ms     *        *     192.0.0.1
  7     *        *        *     Request timed out.
  8    18 ms    19 ms    19 ms  192.0.0.1
  9    19 ms    19 ms    31 ms  10.160.174.201
 10    22 ms    51 ms    19 ms  10.164.176.13
 11    33 ms    33 ms    30 ms  72.14.242.140
 12    31 ms    29 ms    30 ms  108.170.240.193
 13    35 ms    26 ms    35 ms  142.251.76.37
 14    34 ms    33 ms    30 ms  dns.google [8.8.8.8]

 

[gbguy71]

Thanks, while you didn't answer my question about AiMesh (and your comment about mesh is dumb without saying why didn't help). It did point me in a good direction. So, again Thanks!

It would appear that my latency problems lie outside my home network and whether I add a extra internal hop (like using an access point) would have little effect on overall latency.

Here are two separate tracert outputs issued a few minutes across with, to me, problematic latencies noted.

Tracing route to dns.google [8.8.8.8]

1 42 ms 1 ms <1 ms 192.168.252.1
2 74 ms 13 ms 40 ms 104.220.80.1 (Wave/Astound)
3 12 ms 11 ms 11 ms cr1-rhe-a-be153.bb.as11404.net [174.127.183.14]
4 14 ms 21 ms 31 ms be20.cr3-rhe.bb.as11404.net [192.175.29.238]
5 15 ms 16 ms 16 ms be11.cr3-55smarket.bb.as11404.net [192.175.30.212]
6 16 ms 15 ms 18 ms be12.cr4-11greatoaks.bb.as11404.net [192.175.30.222]
7 16 ms 17 ms 22 ms cr1-9greatoaks-be4.bb.as11404.net [192.175.30.216]
8 17 ms 21 ms 17 ms 72.14.222.146
9 17 ms 18 ms 17 ms 142.251.231.97
10 17 ms 16 ms 17 ms 142.251.65.167
11 21 ms 15 ms 15 ms dns.google [8.8.8.8]

Tracing route to dns.google [8.8.8.8]

1 1 ms 1 ms 1 ms 192.168.252.1
2 12 ms 13 ms 12 ms 104.220.80.1 (Wave/Astound)
3 12 ms 16 ms 23 ms cr1-rhe-a-be153.bb.as11404.net [174.127.183.14]
4 12 ms 34 ms 10 ms be20.cr3-rhe.bb.as11404.net [192.175.29.238]
5 18 ms 16 ms 17 ms be11.cr3-55smarket.bb.as11404.net [192.175.30.212]
6 44 ms 38 ms 15 ms be12.cr4-11greatoaks.bb.as11404.net [192.175.30.222]
7 18 ms 16 ms 16 ms cr1-9greatoaks-be4.bb.as11404.net [192.175.30.216]
8 18 ms 24 ms 19 ms 72.14.222.146
9 17 ms 16 ms 18 ms 142.251.231.97
10 18 ms 19 ms 20 ms 142.251.65.167
11 21 ms 41 ms 30 ms dns.google [8.8.8.8]

Trace complete.

Now the question is: Is this normal or not?

 

[Tech Junky]

It could be but, the ISP can tune the routes to be more efficient. The circuits could be a bit saturated and they may already have plans to add more capacity due to those numbers. 5-6 hops to get out of their network isn't too bad it's just the congestion causing latency. I would do a trace during off peak hours and compare. Then send them the traces and ask if they can bring the times down. It also depends on what kind of SLA agreement they have with the circuit vendors. 50ms is about the threshold for most voice apps before you start hearing things. another way to get around some stuff beyond the ISP is VPN as it can speed things up outside of their network with better routing options. Higher bandwidth makes a preferred path and lower latency.

But yeah your network is fine and not the issue.

 

[gbguy71]

Thanks much! Very, very appreciated!