are access point as good as mesh at handoff?

[SpaceTofu]

Hi all,

was considering buying a mesh system, but as I have an Ethernet cable lying already across the house, I thought I'd rather buy an AP than pay money into a new system (where a lot of your money is going into pretty designs and fab marketing!)

I was just curious whether a custom AP would be as good as a mesh system at performing handoff as soon as the signal weakens?

I am thinking of using the modem/router of the ISP to cover the wifi in the front half of the house and then plugging the AP to the cable so to cover the back half.
Currently we are using PoE adapters and I cannot bear any longer with manually switching WiFis and being conscious of where I place them so that IoT devices (smart switches and thermostat) can communicate without issues.
Thanks!

 

[Trip]

was considering buying a mesh system, but as I have an Ethernet cable lying already across the house, I thought I'd rather buy an AP than pay money into a new system (where a lot of your money is going into pretty designs and fab marketing!)

All depends on which "system". Not all mesh systems are just marketing fluff; there are actual products that are flat out superior at producing more optimal handoff events than others, or versus not having centrally-controlled wifi at all. That being said, if you can wire in your APs, you don't necessarily need a true mesh-capable system to produce optimal handoff; that aspect in and of itself is not mesh-dependent. It can be done by non-mesh products, provided every AP uses wire for backhaul.

I was just curious whether a custom AP would be as good as a mesh system at performing handoff as soon as the signal weakens?

By custom, I take it you mean purpose-built? ie. An AP that is designed and built to be that and that alone? If so, yes, of course, but you have to vet the product in question for whether or not it has the designed-in capability to do so, and do so optimally. Easiest way to do that is to look into the datasheets and see what RFCs and 802.11 standards and/or proprietary technologies are covered in the areas of roaming, then cross-reference against actual admin experience.

I am thinking of using the modem/router of the ISP to cover the wifi in the front half of the house and then plugging the AP to the cable so to cover the back half. Currently we are using PoE adapters and I cannot bear any longer with manually switching WiFis and being conscious of where I place them so that IoT devices (smart switches and thermostat) can communicate without issues.
Thanks!

Your proposal of adding a standalone AP to pre-existing standalone wifi will likely not produce the handoff behavior you're looking for. You at least want to be using same-brand APs, running the same firmware version, in WDS at the very least, but that alone only gets you L2 transparency (if it's implemented properly), but doesn't get you the more advanced handoff prompting and handling abilities of a centrally-managed, distribute wifi product. Ultimately, the latter is really what you want if you really aim to ensure the most seamless handoff experience. Examples would be Eero Pro in the consumer space for wired or mesh backhaul, UniFi, Omada or Cisco SB for SMB-grade wired, UniFi Mesh for SMB-grade mesh, or at the top-end, enterprise stuff like Cisco Aironet, Ruckus or Aruba, which more or less handle all desired protocols rock-solidly, and can do both wired or wireless mesh.

Most any system I mentioned above will likely produce better handoff versus just 2 or more standalone APs. Whether the cost difference is valuable enough to you is your call, but hopefully I've helped paint a basic picture of the differences.

 

[SpaceTofu]

All depends on which "system". Not all mesh systems are just marketing fluff; there are actual products that are flat out superior at producing more optimal handoff events than others, or versus not having centrally-controlled wifi at all. That being said, if you can wire in your APs, you don't necessarily need a true mesh-capable system to produce optimal handoff; that aspect in and of itself is not mesh-dependent. It can be done by non-mesh products, provided every AP uses wire for backhaul.

Sorry, didn't mean to that mesh devices are not valid products, they surely are, or certainly some are. However they do come at a price, and I am sure some of that price is being driven by fancier design and lovely advertising.

By custom, I take it you mean purpose-built? ie. An AP that is designed and built to be that and that alone? If so, yes, of course, but you have to vet the product in question for whether or not it has the designed-in capability to do so, and do so optimally. Easiest way to do that is to look into the datasheets and see what RFCs and 802.11 standards and/or proprietary technologies are covered in the areas of roaming, then cross-reference against actual admin experience.

Meant exactly purpose-built, sorry for that. Of course I would look into reviews and feedback, thanks for clarifying. I suppose I am also a victim of the mesh marketing ploy, but if I'm getting it now right, APs and mesh are exactly the same, with the only difference that the network is created over wires in the first case and wirelessly in the latter?

Your proposal of adding a standalone AP to pre-existing standalone wifi will likely not produce the handoff behavior you're looking for. You at least want to be using same-brand APs, running the same firmware version, in WDS at the very least, but that alone only gets you L2 transparency (if it's implemented properly), but doesn't get you the more advanced handoff prompting and handling abilities of a centrally-managed, distribute wifi product. Ultimately, the latter is really what you want if you really aim to ensure the most seamless handoff experience. Examples would be Eero Pro in the consumer space for wired or mesh backhaul, UniFi, Omada or Cisco SB for SMB-grade wired, UniFi Mesh for SMB-grade mesh, or at the top-end, enterprise stuff like Cisco Aironet, Ruckus or Aruba, which more or less handle all desired protocols rock-solidly, and can do both wired or wireless mesh.

Most any system I mentioned above will likely produce better handoff versus just 2 or more standalone APs. Whether the cost difference is valuable enough to you is your call, but hopefully I've helped paint a basic picture of the differences.

OK, now I am getting the impression that actually a mesh solution off the box is better at providing a seamless wireless experience? Regardless of the answer to this, I feel that now that you kindly confirmed my implicit doubt (i.e. that I can not use the wifi of my ISP modem/router in conjunction with one AP), I should probably look into buying a mesh system. I thought they were expensive, but 2 APs don't come at a cheap price tag.
And the mesh system would have the advantage that, if we change house in 5 years time, I can take them with me and easily install them in the new house; or even looking at the present, it could potentially be that where the LAN cable ends is not the most optimal position for the AP for the back of the house. A mesh would offer more flexibility in achieving the best coverage, I suppose.

Damn, I fear I laid for nothing that LAN cable last year!

 

[Trip]

No worries in prefacing what you were looking, or laying down some assumptions. It's all good in my book, just wanted to clarify that seamless handoff is possible with both mech and non-mesh implementations.

As far as choosing a mesh-capable product, the beauty is you can wire in certain APs as root APs (or root nodes) and at the same time deploy other APs wirelessly with mesh backhaul (mesh nodes) whereby the root(s) serve as the priority choice for mesh peering. So your LAN cable is not in vein, presuming it terminates around the spot you'd want to run your other AP.

 

[coxhaus]

I think wireless mesh is going to be good for people with slow internet pipes maybe 100 meg or less. Seems to me 5Ghz wireless will be faster but you need to be real close together. I would guess if you are going to use 2.4Ghz wireless mesh you will be better off using a powerline adapter for wider bandwidth. Of course wire is going to be best.

 

[Klueless]

I'm cheap. If you're happy with your current modem/router it seems far cheaper to simply add a wired AP.

Take a look that the Netgear 6150 for about $75. Linksys, TP, etc offer similar. If it turns out to be the "wrong mistake" you haven't broken the bank. (Do bear in mind that these cheap devices won't likely support "guest" if that's important to you.)

As far as roaming "hand-offs" go at least some of the responsibility rests on the client device.

 

[Trip]

For similar cost, I get it that they aren't new or currently "supported", but there are tons of working-pull Ruckus R500's appearing regularly for $75 on eBay, which when running Unleashed provide probably the best on-the-cheap solution for roaming (that actually works) plus guest wifi. They'll run on PoE (preferred) or 12V/1A AC adapters. The caveats are that you don't mind buying used (they have a very low failure rate) and you've got the skill required for setup, but it's really not rocket science; most anyone who is able to follow written instructions and operate a web browser can have them up and running in under a half-hour. Just a thought.

 

[coxhaus]

So how good is the roaming when doing Wi-Fi calls with Ruckus Trip? My Cisco WAP371 APs drop a few words when they roam.

 

[Trip]

@coxhaus - I roam all the time with my Android phone and tablet and never lose a word, albeit I use the ZoneFlex firmware and a ZD controller, not Unleashed (ZF code base is more mature and a bit more feature-rich). The main thing ZF can do that Unleashed can't is Layer 3 roaming (via tunneling all traffic through the controller), which makes for seamless VoIP while traversing subnets, if your LAN is that complex, otherwise you can choose either firmware. 802.11r and .11k can be enabled in both, and of the several clients for whom I've deployed Unleashed, they all keep calls just fine when roaming, no blips or drops. This of course assumes you've got proper radio density, dB overlap and tuning, plus airtime overhead per AP is kept reasonable and the rest of the network is properly buffered.

EDIT: Removed some potentially confusing lingo.

 

[coxhaus]

@coxhaus - I roam all the time with my Android phone and tablet and never lose a word, albeit I use the ZoneFlex firmware and a ZD controller, not Unleashed (ZF code base is more mature and a bit more feature-rich). The main thing ZF can do that Unleashed can't is Layer 3 roaming (via tunneling all traffic through the controller), which makes for seamless VoIP while traversing subnets, if your LAN is that complex, otherwise you can choose either firmware. 802.11r and .11k can be enabled in both, and of the several clients for whom I've deployed Unleashed, they all keep calls just fine when roaming, no blips or drops. This of course assumes you've got proper radio density, dB overlap and tuning, plus airtime overhead per AP is kept reasonable and the rest of the network is properly buffered.

The one kicker with your stuff is the fact that it's Apple gear, which might benefit most from Aironet, considering Cisco's exclusive collaboration they entered with them a few years ago. For this reason, I know several peers who only deploy Aironet for predominantly iOS and Mac offices. Perhaps worth researching.

I have no Android nor any friends with Androids to test with so I have no answer for that. So does that mean you have to have an Android to roam with a Ruckus for voice calls? I would like to fix my roaming with voice calls on my iphones and losing a few words. It may not be something I can do with the Cisco small business line of wireless APs. I don't know. I don't think I want to go to the Aironet line of Cisco APs as you are getting into the PRO Cisco gear. Costs are pretty high for home use. I sure like the simplicity of the Cisco small business wireless APs where you do not need to run a server to run your controller software. I turned off my server farm a couple of years ago at my home.

Can you give me an example of layer 3 roaming? I have a lot of layer 3 stuff in my home network which I wonder if it is impacting me.

 

[Trip]

It should work just fine with Apple gear; the most meaningful capabilities are largely device and brand agnostic. I probably gave off the wrong impression with my closing statement from the last post, and will edit that out, so as to not confuse others.

For an example L3 roaming reference, see this 10-minute video by Keith Barker (CCIE). The core principles mentioned in the Cisco example are also used by Ruckus, albeit with different nomenclature, but the same behavior is accomplished nonetheless. Notice that at about 5:20 he mentions symmetric tunneling should take <50ms -- by design, so as to preserve VoIP packet flow. Additionally, as this is all device, OS and application agnostic, so it should work as well for Apple stuff as it does for anything else.

All that being said, the benefits here are obviously only applicable if you are running your APs on different subnets, which, even though you have L3 gear in your LAN, may nor may not be the case. If not, Ruckus still may be worth a try, given that the baseline roaming implementation itself could likely produce a smoother packet flow than the Cisco SB code. Two R500's is all you'd need to find out. I'd venture they'd at least get you comparable endpoint performance out at the edges. If the baseline test works then you could add a third. Might be worth a shot.

Hope that helps.

 

[coxhaus]

Well that tells me a lot as there is nothing in my small layer 3 network that takes over 50ms to reach. I should be good to go. I don't know why it drops a few words on roaming.

I still don't know if iPhone voice call roaming works prefect with Ruckus. I see should and will as there is still a question.

 

[Trip]

It's not so much that a LAN would have 50ms of transit to any of its parts, moreover than a roam event betwen subnets, if you have two or more, could take much longer, if L3 roaming wasn't implemented. Since it doesn't sounds like you have multiple subnets, then then the feature of course doesn't apply, and has no relevance to your issues. Just wanted to give you that example so you were aware of what it did offer, if it could be useful.

Looking back for a moment at clients, if you haven't already, perhaps do a test call and run a packet-capture (if that's possible?) on the iPad/iPhone while roaming, then analyze and see if you can ascertain whether it's the endpoint causing the interrupt.

Another thought - Perhaps iOS's tendency to be a sticky client could be combining with perhaps an imperfect balance of perimeter broadcast strength between APs? Most things I've read state -65 to -60 dB as the optimal balance of signal strength to roaming allowance. I would presume your coverage is solid with good relative RSSI, but figured I'd ask to be sure.

 

[SpaceTofu]

No worries in prefacing what you were looking, or laying down some assumptions. It's all good in my book, just wanted to clarify that seamless handoff is possible with both mech and non-mesh implementations.

As far as choosing a mesh-capable product, the beauty is you can wire in certain APs as root APs (or root nodes) and at the same time deploy other APs wirelessly with mesh backhaul (mesh nodes) whereby the root(s) serve as the priority choice for mesh peering. So your LAN cable is not in vein, presuming it terminates around the spot you'd want to run your other AP.

Thank you.
If I interpret this correctly, you are suggesting to plug the root mesh AP where I would plug my bog standard wired AP (in blue) and then have one mesh node where the current position of my router is?

This one above is a diagram of my situation, hail M$ Paint!

But then I would still be forced to buy a 2-set mesh AP, which is quite dear.

I don't have extreme requirements, nor am I particularly needy with regards to all the tweaks and setups that can be changed in a net product; but I can play with it if I have to, and to save some money I feel like I have to.

I tried to read and understand all the following messages, but unfortunately I didn't take home much from those.

How would a Ruckus be useful in my scenario? Does it excel particularly at handoffs and can be integrated with any bog standard router for a seamless Wifi experience?

Thank you for the help

 

[Trip]

@SpaceTofu - Considering you have wired ports in both places (a LAN port on the router itself on Floor 1, and an ethernet cable going to the AP position on Floor 2), I would simply wired in both APs. This will allow you to run a simpler, wire-first/wire-only product, as opposed to requiring something that supports wireless backhaul (ie. "mesh"). In regards to handoof, I've have read numerous good things about TP-Link Omada, provided you include the $85 OC200 controller on the network. Couple that with two $65 EAP225v3's or $97 EAP245's and that combo would stand a chance of doing quite well. I also know that @coxhaus has had mostly positive experiences with Cisco WAPs; they can run clustered, no controller required, and from most of what I gather, seem to do a decent enough job at handoff as well (YMMV).

As far as Ruckus or anything enterprise-class goes, we have to again remember that it's the endpoint that determines most of the roam event and subsequent behavior, but I can tell you I've never personally had, nor have any of my clients really ever had, a problem with seamless roaming using Ruckus gear. I suspect this is due to better implementation of all roaming-related standards in their code base, but I haven't gone into lengthy-enough testing or research versus peer products or lower-end gear to confirm.

For this project, enterprise would probably be overkill if you're a bit put off by the stuff in general and/or buying refurb/working-pull hardware; totally understandable on both counts.

 

[coxhaus]

It's not so much that a LAN would have 50ms of transit to any of its parts, moreover than a roam event betwen subnets, if you have two or more, could take much longer, if L3 roaming wasn't implemented. Since it doesn't sounds like you have multiple subnets, then then the feature of course doesn't apply, and has no relevance to your issues. Just wanted to give you that example so you were aware of what it did offer, if it could be useful.

Looking back for a moment at clients, if you haven't already, perhaps do a test call and run a packet-capture (if that's possible?) on the iPad/iPhone while roaming, then analyze and see if you can ascertain whether it's the endpoint causing the interrupt.

Another thought - Perhaps iOS's tendency to be a sticky client could be combining with perhaps an imperfect balance of perimeter broadcast strength between APs? Most things I've read state -65 to -60 dB as the optimal balance of signal strength to roaming allowance. I would presume your coverage is solid with good relative RSSI, but figured I'd ask to be sure.

I do have multiple subnets running. If fact my Cisco wireless device is setup on a trunk port with multiple VLANs. The VLANs are different subnets. But each VLAN which is a separate subnet or network and it is tied to a SSID so my roaming is only across 1 SSID which makes my roaming within 1 subnet. Now that I think about it I am only doing layer 2 roaming per SSID.

My Cisco WAP371 APs are on the way out and I have new Cisco WAP581 APs coming. I will wait to trouble shoot the new APs if they don't work. The technology is several years newer in Cisco WAP581 APs.

My guess is the client is not sticking but a couple packets are being dropped as you roam. It is like a part of a sentence is missing. FaceTime roaming works fine with no dropped calls.

 

[SpaceTofu]

I suppose Ruckus is out of the question. I did enjoy (probably more than a decade ago) playing around with net stuff (didn't DD WRT bring a lot of joy to my back then life!), but right now I would like something that works quite decently out of the box and only requires a little bit of fiddling about to make it work great.

Granted that we assessed that I need two new APs (as opposed to one AP and the already owned modem / wirless router), do I actually really need the TP-Link Omada OC200 controller? It is a fabulous piece of kit, but again, probably unnecessary for my needs. Is it needed at all from a technical standpoint?
If yes. then I am starting to think I should buy a mesh system, as purchasing a OC200 + 2 APs = $85 + $130 = $ 215
Meanwhile a mesh system (Orbi) is "only " $ 280 (in UK at least). With a mesh system I would have the added flexibility that if we change home, I could still use it, meanwhile the wired APs solution would probably needed to be sold.

To really sum it up ( sorry if I am going round in circles), I think I need to understand whether the OC200 would give me any additional benefit, as if I can shave off those $ 80 from the total bill, then I am sure a wired APs is the way to go!

 

[Trip]

Regarding Omada, yes, the OC200 controller is required to in order to ensure 802.11r/k are administered properly. So it's a must for anyone looking for the "seamless" roaming experience. That said, a 2-unit Orbi kit would likely also suffice, and be simpler to setup. I'm not sure as to how well they've developed seamless roaming, however; last I checked via their forums last year, it was a still a bit half-baked, so you want to do your due-diligence there.

Also, do be aware that if you stand a chance of using Orbi via mesh in the future, it's only layout option is a "star" topology, where each satellite mesh node must backhaul directly to the root node (thus forming a "star", sometimes called "hub and spoke"). This is in contrast to true multi-point mesh, or multi-level mesh, where wireless nodes can link to each other at two, three and more hops away from a wired root node. Multi-level is supported by Eero Pro, UniFi Mesh, or most any enterprise product; obviously cost will be higher, and compounded further with that many active nodes, so not really relevant to your setup now, but just wanted to make you aware of the difference in capability.

That said, even with the topology limitation, Orbi will of course work fine for your setup as it is now. Plus, by the time you'd ever look into adding satellites and/or redeploying at a new residence, there will probably be a more effective product out there to replace the whole system at similar pricing.

 

[SpaceTofu]

Thank you for the added info, @Trip, haven't I learned a lot of valuable information in this thread!

To make sure I understand this 100% correctly: multi-level mesh would give me the ability (or luxury!) of having, say, three very distinct WiFi SSIDs in my home, one for private use, one for guests, one for IoT devices?

 

[Trip]

No, multi-level, or more accurately, multi-point, mesh has nothing to do with how many SSID's can be broadcast by the system at any one time. Multi-point refers to the link capabilities of the AP-to-AP connections, and consequently the type of topology (or mapped layout) of APs that can be setup. Here is a picture to help:

In the graphic above, each gray circle would be an AP, and each double-ended blue arrow would be an AP-to-AP wireless backhaul link. Orbi would be able to do "Point-to-Point", "Star" or "Point to Multi-Point". The system is designed to operated with one and only one "base" (or root) AP. True multi-point systems can operate with multiple root APs (such as UniFi Mesh or Ruckus) and can thus also run in "Fixed Mesh" and/or "Cluster" configurations, per above. Make sense now?