I just see allot of posts of users using a multiple hubs. What are the criteria for determining you need a second hub. Is it because of the size of a mesh or number of devices, or is simply to manage different types connection types. I am just still not very clear as to why or when I would start to consider a second hub. I haven't noticed any issues with my current setup, this is mostly out of curiosity then anything else.
I think the official response is you should not need a second hub.
If you are experiencing slow downs you should attempt troubleshooting and try to identify / correct the cause.
Unofficially some of us find separating our stuff across hubs helps mitigate slow downs. AFAIK there is no official criteria for when you should consider multiple hubs.
2 Likes
A common use for a second hub is if one has an 'out-building' that is too far away for a Zigbee or Z-Wave mesh network to reach. If one has Ethernet available in the out-building, then a second Hub can easily create a separate Zigbee and Z-wave mesh, which can then be integrated with the main house via Hubitat's built in "Hub Mesh" software, to share devices between hubs.
3 Likes
Generally speaking you shouldn't need one.. but like @cwwilson08 and @ogiewon say some of us have used multiple hubs to mitigate certain issues like with slowness and distance. It really depends on your use-case.
I have installed a single C-7 at a client's large older (late 1800s) house and with a judicious use of repeaters etc things have been working very well. Unfortunately to @ogiewon's point - my client's detached garage was just too far away to be reliable but they had an ethernet run to it so was able to add a spare C-5 I had on hand. This has worked out great so far.
Some other reasons:
- To isolate certain troublesome devices from the rest of the mesh.
- To mitigate failure modes and help with recovery - if you have 2 hubs in different areas of the house if one fails the other keeps working, depending on how you have your rules etc set up.
- Large numbers of devices and rules. While not strictly necessary I've found based on experience with my clients that reducing overhead by limiting amount of devices per hub (150 or less maybe) can really help. At least I seem to get less support calls.
4 Likes
This is actually a common use. For a long time, I had a separate C-5 that had Xiaomi Aqara/Mijia compatible repeaters and Aqara/Mijia sensors. Some people also find it convenient to put zigbee bulbs that are repeaters (Cree, GE Link, Ecosmart, Sylvania/Osram) on a separate hub.
Currently, I have all my z-wave devices on a C-7, and all my zigbee devices on a C-5.
1 Like
That's the way I have mine set up as well - C-5 for Zigbee (I think the hub has the same ZB radio as the C-7) and C-7 for the Z-Wave devices.
I used to have 2 C-4's by location - one upstairs and one on my main floor. This configuration worked really well too. I did this during a prior period where hubs where experiencing terrible slowdowns (much worse than todays issues) and rebooter apps became a popular thing.. Maintenance was much easier. Could bring down one hub while the other kept running etc.
1 Like
For now, I have left any devices that pair with S0 on my old, original, C4. Also have left two Zigbee devices that I have to migrate but they are in ceiling fan shrouds and I have been too lazy to get up there (to push the button) so I can move them to the C7 with the other Zigbee devices.
I want the lowest latency on automated lighting routines so I have my Zigbee motion sensors (which is most of them) and the Lutron Caseta integration on the same C7.
The Z-Wave devices that pair nicely are on another C7 that will (hopefully) eventually be home to all the Z-Wave.
2 Likes
So to summarize.
- There is no real criteria for when one should look at additional hubs.
- Some times they are used when physical distance is to great to overcome with repeaters
- To hold/keep problem devices from interfering with other devices or generations of tech. ie keep new stuff with new stuff old stuff with old stuff.
- Potential soft limit is around 150 devices that is not documented.
2 Likes
Yep - just based on your experience and use-case.
Possibly but also for ease of maintenance and rule overhead reduction. Note you still need the satellite hub to have a connection to the network either via ethernet, wifi, moca-ethernet, or powerline-ethernet.
Yep that works too! Sometimes it's isolating tech from tech as well not just the old stuff.
That is my assessment only as a somewhat skittish installer not the general consensus. There are people in this community who have 200+ devices and are running fine and there are people who have complained after 60-70 devices.
if you're taxing your primary hub then it might be time for a second one. i only have 2 hubs because of the upgraded model, but have also turned it into a test hub before i do stuff on my primary hub. this was helpful when testing apps to see how it impacts the hub. it has also become a different protocol hub, so i have z-wave on one hub and zigbee on another. i could've easily put both protocols on one hub, but i was too lazy to exclude and include again.
1 Like
I have 4 hubs interconnected. I chose a 'location' based topology.
1 hub is upstairs and is ZWave and my only Zigbee.
1 hub is downstairs and is ZWave.
1 hub is for Internet facing components.. Alexa, Google Home, Siri/HomeKit, Dashboards and Weather, etc.
My latest hub is labeled "Front" to remind me I've created another area for the C-7 (ZWave only) to allow me to glacially migrate 'downstairs' to 'front'. 
I did it initially to just see if I could. Clearly the answer is "Yes, I can" and it hasn't stopped. It's become, for me, a possibility of a shorter recovery from a hub hardware failure. Haven't had one yet, but every minute is a new opportunity.
I also like applying upgrades and not risk the whole house.
I looked at the price I'd paid in ZWave (or Zigbee) devices and realized that the cost of another hub was relatively small. A hub is $129 today or about 4 Inovelli Wall Switches/Dimmers? Clearly for me, I want a lot of reliability by splitting my home into 'tiny homes' each with it's own hub controlling a smallish number of devices.
1 Like
I have 259 devices on a single C5 hub with no issues. I do have a second hub, but it's dedicated to my problematic GE Link bulbs.
4 Likes
See I wasn't really asking this question because I am having issues or think I need to, but instead to just start to prep myself incase so I didn't just hit a wall and struggle for a while until this was the conclusion to come to. I wold rather try to be a little forward thinking so i can prep myself and be thinking about this as I approach a given limit and be ready to mitigate you know. I have 91 devices on my device screen with a handful of them being virtual or shared cloud devices from Smartthings with Hubconnect. So though i am not experiencing problems, it is something that isn't without some relevance to be concerned with.
I kind of look at this as a given, but how do you know when you are actually taxing a hub. I have a node red hub running on my Unraid home server collecting data. I guess that will provide the clue when timings start to rise or cpu temps stays high from lack of idle time.
All devices are not made equal either. You could have a bad zwave device kill the radio or be added in a way that slams it, or maybe have a poorly written driver hitting a wifi api that causes a ton of load. So this really is a unique item for each user is really the take away.
In my summary above 2 and 3 are fairly straight forward, but 4 with a number of device based on hub load is a complete wild card it seems.
Perhaps the better question would be under conditions of hub load what are the objective methods for discerning when a second hub is needed. Is it simply response time.
If you are running Node-RED why not move your rules over too? That helps a lot - at least it did in my case.
In my case it is simply a matter of having time to dig into node red and get familiar with it. I have Node Red, Home Assistant, Graphna, and Influx running on my Unraid server which is a old Core i5 2400 with 24GB of ram. It is a good little home server. Node Red and Home Assistant were loaded for the sake of dabbling with them, and getting to understand what all the fuss is about. I don't have anything in Hass.io setup, and one flow in node red that was distributed here. It is performance flow that writes hub details to InfluxDB, and then i modified it to pull the recent addition of Temps reported on the hub to InfluxDB.
One thing i am cautious about with Node Red is that i believe it is basically unaware of state. There are ways i see that being a problem. I guess of course that may just mean write the flows to accommodate for that, but i need a better understanding of node red before i go crazy with it. and i need time to get that.
I'm not sure what you mean by being unaware of state? HE Nodes maintain device state in a cache and (can) rebuild on hub restart. Any events get sent via Maker API/webhook and recorded in the cache.
Here's a very simple pico control sequence for my den lights..
Note: the status below the "Den Pico" device node is showing the last button pushed.. Also note that the "Den Lights" device node returns it's state value and I use that to do the toggle...
Well like I said I need a better understanding of it. I could be completely wrong with my impression as of right now.
1 Like
This may be one of the best reasons for multiple hubs. The radios are serial devices which means one misbehaving device can cause havoc for the hub with which it is communicating. Spreading that out, seems to me, is a strategy for avoiding a widespread breakdown or slowdown.
Well it is and it isn't right. All you do is mask or hide the problem potentially by splitting it off and only impacting a small portion of the overall environment. In some cases wouldn't it be better to identify, replace and eliminate that hardware if it is problematic.
** i know this may also be a personal view as well. I just see so often as a IT professional where we just through hardware at a issue instead of fixing the problem. That is kind of like what that sounds like. Of course you also don't want to create problem where a issue will bring down a service so yea i understand that to.
1 Like
I totally agree but sometimes it is hard to chase them down. Over the years I have had a couple of motion sensors that were driving me crazy until I found them (and threw them in the trashcan).
