MESH Linked EAP225_Outdoor Kinda Slow
MESH Linked EAP225_Outdoor Kinda Slow
So,
I've been trying to learn to love my relatively new EAP225-Outdoor that I have had installed for about 2 months now. I have 2xEAP225 indoor and an OC-200 that I dearly love, but the Outdoor unit has always seemed "Meh" at best. I've seen a lot of great reviews on Amazon, but my experiences haven't been that great.
Don't get me wrong. It's works, my OC-200 adopted it and configured it right away, and it seems, so far, just as stable as my indoor EAP's. It works well with my OC-200, configuration, status, fast roaming, etc. All just as good as my regular EAP's.
But speed wise, signal strength wise, and performance consistency it's never really worked as well as I was expecting. Now to be fair, it is on wireless MESH, which I also don't have much experience with. I'm sure if I could run an Ethernet line to it, it would work much better.
So today I spent some time optimzing it's placement and configuration, and I've definately made some improvements. Particularly in terms of signal strenth, signal consistency, and and throughput DL/UL consistency. But it's still slower than I would have expected.
Here are the changes I've made that have helped:
1) Moved it further from my house (and thus further away from the master node). It used to be about 10-15 feet away from the AP through an outdoor wall (drywall, insulation, vinyl siding) mounted on of my Deck posts. It's now at the edge of my back yard, probably ~50 feet away.
2) I've lowered the center of the antenna height from ~8 feet to ~5 feet -- the 3 or 4 dBi antenna should be pretty omnidirectional, but the unit definately seems to work better with the antenna at device height.
I suspect it was a combination of being too close to it with the antenna to high. It just so happened I was bouncing between one of the minor lobes of the RF signal. I tried removing the antennas to see if they were doing anything, and the AP quickly lost it's uplink connection (which works fine when they are on) so at least one of the antennas is working. I've ordered some replacement antennas which may or may not be higher quality, 3 dBi and 6 dBi to experiment with.
3) I changed from using High / Medium transmit power to using the Custom scale. I don't know why by the signal seems a lot strong on Custom -- even when setting the equivalent dBm (e.g. 14 or 22) which corresponds to Medium and High. 14 doesn't seem to work that well. But 16 and 18 are very good. I'm on 18 now.
4) I turned off the 2.4GHz radio on the Outdoor unit. I don't need it, and I'd rather reduce the traffic and interference. It also helps me keep so few random indoor STA's from occassionally connecting to the Outdoor AP mistakenly.
So, between all of these changes the Outdoor unit now provides good 5.8GHz coverage for my entire back yard (mostly 3 bars on my iPhone, sometimes 2) -- roughly 1/4 acre -- 100 feet x 50 feet. The AP is centered on my back fence (which is 100 feet long) so it's now providing a semicircular coverage area with a radius of 50 feet (to the house) and a bit of coverage to my side yards (on either side of the house) about 60-75 feet away.
Speeds are even and predictable now, with good mobile client STA RSSI's, but speeds slower than I would have expected. I'm only seeing 140-150 Mbit/sec of "good put," DL and UL, best case, using various Internet speed tests (I have Gigabit FIOS). In contrast I see speeds of 350-450 Mbit on my indoor, hardwired AP's.
If you look at the image below, you'll see that signal from the root node is good (-68 dBm) and the air link rates are high (650 Mbit/sec up and down). So based on my normal calculations, I would expect / hope to see roughly 50% of the air link rate as usable "good put." That is to say ~325 bit/sec. When in practice I'm only seeing half of this at best -- e.g. ~162 Mbit/s of usable throughput.
It seems like the wireless MESH is "dinging me twice." That is to say, I'm only getting roughly 1/4th of the air link speed as useable throughput. Or perhaps more precisely if there is a 50% loss/overhead from STA to Outdoor AP, and then another 50% loss/overhead from Outdoor AP to root node. That would be 0.5 * 0.5 = 0.25 of useable BW which is what I'm seeing.
While I expected some additional wireless MESH overhead, I wasn't (perhaps naively) expecting this much.
Any suggestions?
- Copy Link
- Subscribe
- Bookmark
- Report Inappropriate Content
@Doc2485, I also agree that wireless links between APs should always be avoided in business installations. But there are situations where you just can't wire up APs, e.g. if you need to deploy a WLAN in listed buildings (historic buildings), which require permission from authorities to even drill a hole in a wall or a ceiling for a cable duct. Much paperwork and at the end it can be that authorities don't permit it at all.
@JSchnee21, the goodput you can reach over a wired EAP is already 50% (a/b modes) or 70% (g/n/ac modes) of the WiFi rate. The other 50% loss of throughput comes from the wireless mesh link, so yes, you're right that a first-hop mesh node will only achieve a ¼ of the average of both WiFi rates between both meshed EAPs and between the last EAP and the client.
What's more, the mesh network is subject to the Hidden Node Problem if clients share the 5 GHz mesh backhaul. Thus, turning off the 2.4 GHz band of a mesh node can increase interferences in the 5 GHz band.
- Copy Link
- Report Inappropriate Content
Hi R1D2,
Could you clarify this:
"What's more, the mesh network is subject to the Hidden Node Problem if clients share the 5 GHz mesh backhaul. Thus, turning off the 2.4 GHz band of a mesh node can increase interferences in the 5 GHz band."
Ideally, I would think that it would be better to have clients on the 2.4G radio rather than the 5G so that the 5G was "free" for backhaul. I think that is what you are saying. That is, if you host an SSID on both radios (with band steering) but turn off the 2.4G, then clients will be shifted to the 5G and compete for airtime with the backhaul.
This was my recommendation to ByteGuy recently (put all STA's on 2.4G and leave 5G free for back haul) given his extensive MESH setup. Especially since each STA is throttled / BW limited to 10Mbit/sec DL & UL anyway.
But, I'm not sure if my suggestion was ever fully implemented.
-Jonathan
- Copy Link
- Report Inappropriate Content
JSchnee21 wrote
Could you clarify this:
"What's more, the mesh network is subject to the Hidden Node Problem if clients share the 5 GHz mesh backhaul. Thus, turning off the 2.4 GHz band of a mesh node can increase interferences in the 5 GHz band."
All devices always listen to wireless signals in order to detect whether some device has successfully acquired the channel for sending data. That's part of the standard Carrier Sense Multiple Access/Collision Avoidance (CSMA/CA) mechanism. As long as they detect a signal, they will pause.
Now, if the root node wants to send data to the mesh node, it waits for silence, acquires the channel and sends data to the mesh node, which in turn will send it to the client. The root node still can detect that the mesh node is currently using the channel to send data to the client.
But if the client replies and the root node cannot receive the client's signal, it could falsely detect silence, send more data to the mesh node and collisions / interferences will occur at the mesh node which now receives data from two devices, the client and the root node.
This is called the Hidden Node Problem, b/c the client is hidden to the root node. The HNP is what makes repeaters, which always use the same channel as the AP whose signal is to be repeated, so unreliable.
Ideally, I would think that it would be better to have clients on the 2.4G radio rather than the 5G so that the 5G was "free" for backhaul. I think that is what you are saying. That is, if you host an SSID on both radios (with band steering) but turn off the 2.4G, then clients will be shifted to the 5G and compete for airtime with the backhaul.
That's the solution to avoid the HNP. If the 2.4 GHz SSID or radio is turned off, clients will connect to the 5 GHz band regardless of the band steering setting in Omada controller.
Note that interferences (resp. collisions) don't prevent communication at all, but goodput will be decreased significantly if interferences reach a certain level. As long as there is enough AirTime, collisions don't harm. Collisions are still detected – even a sender listens to what it sends and thus can verify the data transmission –, but better than detecting it is to avoid collisions at first.
This was my recommendation to ByteGuy recently (put all STA's on 2.4G and leave 5G free for back haul) given his extensive MESH setup. Especially since each STA is throttled / BW limited to 10Mbit/sec DL & UL anyway.
Yes, you're correct (as always :-) ).
If we deploy a mesh, say, in a hotel, it's often b/c we need to extend WLAN to a room which is badly covered by the main EAP. In this case (only few client devices expected) we enable both frequency bands for clients. But if the meshed EAP should serve a whole floor and total goodput decreases during heavy use, it can help to enable just the 2.4 GHz band for clients and use the 5 GHz channel only for the backhaul.
- Copy Link
- Report Inappropriate Content
R1D2,
Thank you for the sanity check. Makes sense.
I cannot wait for an EAP with the new Qualcomm 1210 Wifi-6E platform -- 4x4x4 (2.4G, 5G, and 6G). Or even the 810 with 2x2x4. 4 stream, 160MHz, on 6GHz will make for some sweet wireless backhaul.
doubleu, doubleu, doubleu dot snbforums.com/threads/qualcomm-rolls-out-wi-fi-6e-lineup.64418/
Do you have 6E in the UK/EU?
-Jonathan
- Copy Link
- Report Inappropriate Content
JSchnee21 wrote
Do you have 6E in the UK/EU?
Authorities here still work on EU-wide rules. They are not the fastet if it comes to real work.
- Copy Link
- Report Inappropriate Content
Ha Ha Ha!
My employer has a large site in Belgium and we work with them routinely. The running joke in our office is that everytime we meet with them to "just get something done" they always want to setup even more meetings, have more discussion, and make more PowerPoints!
The American / Nike concept of "Just Do It" is totally lost on them. Often our needs, requirements, priorities change so fast that if you plan and pontificate for too long you'll miss the opportunity to respond in time.
-Jonathan
- Copy Link
- Report Inappropriate Content
JSchnee21 wrote
The running joke in our office is that everytime we meet with them to "just get something done" they always want to setup even more meetings, have more discussion, and make more PowerPoints!
To just get something done in the EU, you need to fill in up to 15kg declarations of compliance with EU rules on paper and have to wait 3-5 years until the thing you want to do would principially get approved if there were not the new rules from last year, which changed certain regulations again. :-)
- Copy Link
- Report Inappropriate Content
Information
Helpful: 0
Views: 2428
Replies: 17
Voters 0
No one has voted for it yet.