OpenHAB Server Appliance

Hi All,

I am interested in building an OpenHAB Server Appliance that would be able to control Schlage Doorlocks and Lutron or Leviton Light Switches. I’d like to include Z-Wave and WiFi capability into the appliance. If possible, i’d like to add other connectivity types.

One other important fact. I want to test this appliance for large installations, in buildings about 60,000 sq. ft. in size. Not an open area like a warehouse, but more like an office building.

Any ideas how to do something like that?



Hello Jason

This project sounds interessing. Let me give you some inputs, as good as I can.

At first you should know something about your SLA. How long could these systems be offline before it impacts the business?
How many items should be controlled from this appliance? Did you know, zWave has a limited amount of manageable devices.
How many rules should run?

With this information, you can start to plan your hardware…
How about redundancy to fulfill your SLA?
Did you know, only a few zWave Sticks could be backed up? At least this is your single point of failure.

If you use zWave in such a wide area, the planing is essential. zWave builds a mesh-network (every wired device act also as a repeater), but some devices should communicate with their controller directly.

If you can provide further information, we could give you more advice.


Great questions Michael. This whole project is really an experiment to test OpenHAB in a large setting. To see how viable it can be, and what the challenges might be. As you point out the SLA is on the top of my list.

The test site will most probably be a school, with a zWave lightswitch in each classroom, and some type of smart lock deployed at main doors and potentially classrooms. So, lets say, 50 classrooms, and 50 locks, spread out every 25’ over a 60k sq ft building. I am aware that zWave has a 258? device limit, and that it works as a mesh. But I am not limited to zWave. There will be wifi throughout the building as well. What I want to stay away from is expensive wiring, otherwise my budget will not be approved. If power goes out, the only thing that needs to work are the OpenHAB Server and the main building locks (which are on battery). So that should be ok, since they operate on WiFi.

You should know that the SECURITY command class is experimental at best and certainly not complete nor reliable yet. See the following thread for more details:

There are other ways around this such as getting a Vera or other hub that does implement this command class and have OH interact with it.

Also, even when/if it is fully supported, it means that all your locks must be in range of your controller and your controller must be plugged into your OH server in order to achieve inclusion into the network. This could be a significant problem in large setting like this. At a minimum it means you either need to pair the locks with the controller before installing on the doors, or install OH and the zwave dongle on a computer that can be taken to the doors themselves.

I don’t know if any other technology locks are supported. I know Chris has been adding lots of new capabilities to the OH 2 Bluetooth binding and the security on a lot of WiFi locks is poor enough there may be ways to control them from OH as well using the HTTP or TCP binding.


Controlling more than 232 devices needs multiple controllers, especially for extended range. Z-Wave is a mesh network so devices can forward messages to distant kit up to 4 hops, with the official docs estimating each hop at up to 40m giving a total of ~200m:

Your school example could use multiple OpenHAB nodes (Raspberry Pi 2?) and conventional Z-Wave USB controllers, but for an extreme example referenced in a giant hotel a different architecture was referenced.

A few months ago Sigma announced a new Z-Wave to Ethernet gateway called the ZIPR:

I can’t find the article now that linked this gateway device to a very large Z-Wave deployment in the The Wynn Hotel & Casino, Las Vegas, however the suggestion is each room has a ZIPR forming a separate Z-Wave network (assume 32bit Home-ID), allowing the full 232 nodes per room with the local RF to Ethernet interface communicating back to a pool of central servers:

The catch is that until last week, use of the ZIPR needed a NDA to access the documentation and API for Z-Wave over IP or Z/IP. That has partially changed, with Sigma Designs releasing the APIs and basic dev tools.

The ‘official’ Sigma Designs tools appear to include binary blobs available only for specific platforms, which makes their drivers of little direct use (apart from testing perhaps) to OpenHAB which uses Java for full platform independence.

In theory, the opening of the documentation does mean a driver could be created for Z/IP but my understanding is the priority is to improve the core OpenHAB2 platform and the Z-Wave bindings to drive conventional serial interfaces via USB or serial. This includes the security class used by locks, and other newer devices.

To manage the risk of a large installation with public safety considerations, personally I’d want to do a lot of testing before moving away from conventional hard-wired access control systems using (say) magnetic locks. Wireless systems are very easy to install, but you need to consider the impact and costs of replacing batteries to fix an unexpected failure at a peak time!

This is all great information. I think I will start off with the main Administrative area of the building, and grow the network from there. I should have at least one zwave node every 20 feet. The Yale Zwave locks look like a good solution for that problem also. I’ll report back on the status once I have it operational.

One other thing to consider with zwave. Only mains powered devices will act as a relay. So your deadbolts will not forward messages it receives from neighbors and therefore do nothing to improve the quality of the mesh. Assuming you have a mains powered switch in each room you should be fine, but this is a limitation you should be aware of.