RE: Advice on replacement for Elve

Yup; here a few years back tested Homeseer running in a VM using using a Digi USB Anywhere and Lantronix USB device talking two two Digi 8 port usb to serial devices for some total of 16 serial plus 5 USB devices.  Two of the USB devices carried audio; one via a Way2Call box and another using a  USB sound card.
 
It worked well and I was pleased.  Today there are a number of Homeseer users running Homeseer on a VM. 
 
I'm also sitting here working in VMware right now anyway so I just double-checked, and you can hard-set the MAC address if you wish - then it should never change.
 
Thanks Work2Play we do that for some machines I have just found it to be a little unreliable.
We hard-set a VM MAC but the license was still having issues, I'm not sure if the license was looking at something other than the MAC address too.
You are right though I would much rather use VMware. The VM did work very well for Elve and I was more than happy to use it but I just could not trust it with their license setup.
Thanks Pete_c I have never had any performance issues with VM's either.
Just have to convince my bosses to spend the $$ to replace Elve.
 
Work2Play said:
Even migrating to new hardware as things change or upgrading or whatever - it's done just by moving the VM around.
 
Sure, the trouble is VMs don't get real time access to hardware.  But for a lot of other stuff the software shims between them and the actual hardware do a fair bit of the proxying.  But for real time stuff, you can't fake that.  Now, it's entirely possible some might tolerate having a control system not be able to respond in real time.  Portabilty of a VM seems like a fine idea until you start counting the performance hit that comes along with it.  For some stuff they're great, but for this sort of thing I'd prefer a standalone machine.
 
As for licensing, some stuff won't even run if it detects it's in a VM.  Too easy to hack the licensing/DRM from the hypervisor ring.  But then licensing is a whole other set of arguments...
 
What "real time stuff" are you worried about? I can't think of anything in HA software that needs to happen any faster than the milliseconds range. VMs can certainly respond to any task in this time frame. Serial, USB and Ethernet communications have all been shown to work fine under VMs.
 
I run CQC in a VM and so far the only issues I have had were not CQC related.   They mostly centered around getting COM ports and USB devices to pass through successfully to the guest.  But I am running on the now deprecated VMWare Server not on an ESX host...   I also use RDP as my primary means of accessing CQC through android phones/tablets and there is no issues with delay, etc (as long as you assign enough resources).  But you can run the interface viewer on any windows machine, or use a RIVA client on IOS or android clients.
 
CQC itself is very lightweight, I originally started running it in a 512MB VM on XP.  I have since bumped that up to 2GB in order to support multiple RDP sessions.    
 
It is modular (as I think Elve was) so you can distribute the various components to run on machines near your resources,  although they may need to all be in the same subnet/broadcast domain to find each other.  Jonathan was a long time user and big contributor to CQC before founding Elve, so a lot of the concepts are similar.  The primary fundamental difference that I see is Elve leverages .NET as something more people are comfortable with, whereas CQC tries not to be dependent on underlying libraries in an effort to maintain stability.   So it's all custom windows/dialogs, macro language, etc.
 
Yup here moving in the direction away from using Wintel as an OS for my automation. (the faster the better for me). 
 
I have also tested my CCTV stuff in a VM and while it works still prefer my little hybrid combo of an 8 port capture card plus a mixture of IP camera capturing in a box rather than a VM.  Same with my MythTV box now talking to a few tuners (just added an old analog tuner for in home TV channels).  (I came from the old Tivo world from many years ago).
 
Nothing against Wintel here but found that a simple Linux basic OS running on a tiny or small almost embedded linux appliance works best for me.
 
Guessing that the HAI OPII stuff / methodology has had an effect on me over the years....
 
If this makes any sense; its just the meat with very little garnishes of the fat of an OS.
 
That said also doing this now with my touchscreens.  I did originally build a tiny XP hodgepodge of an almost embedded XP but a kludge of a mixture of tablet and media center, text to speech and other stuff in less than 2Gb (really even smaller) of an SSD flash drive. (but that is 16Gb in space). 
 
Playing with a pure linux automation touchscreen now...initially building an XBMC propietary OS running on MMC memory.  It worked but DB used up the rest of the 2Gb.  That said went to an external network db and that worked.  Today the Linux touchscreen is running squeezelite, xbmc and HSTouch in a very tight almost propietary linux build.  Personally liking this better than Android....but that is me....don't really care for the Android "fluff"...that is personal though...don't need it for home automation...works fine for phones though....
 
I am though comparing this to a very tight WIndows 8.1 build (< 4Gb "lite" version) which is very tablet like...but appears to be a bit fatty still....liking it a bit for tabletop tablet use....the guts of the hardware lets me run Android, Linux or Wintel such that I can compare my touchscreen features; likes and dislikes between the different OS's and provides me with the ability to remote the device from the mothership....
 

Attachments

  • pic-4.jpg
    pic-4.jpg
    121.6 KB · Views: 13
  • pic-5.jpg
    pic-5.jpg
    175.8 KB · Views: 12
We ( previous gig ) used to use the phrases "soft real time" and "hard real time" to try to draw a distinction between classes of applications.
 
While it was a moving boundary hard real time generally applied to high speed instrumentation, sampling and such. soft real time we applied to 'human organ perceivable' time scale rates.
 
What's great about enterprise VMs is how easy it is to have a hot spare.
 
What someone might want to do in a residence is one thing.  What someone might want in an enterprise might need is perhaps entire different. 
 
My opinion that a system integrating 40 different Elks across an entire campus might not benefit from being 'just another VM' on the server.  Sure, there's lots to be said for bulking up a lot of stuff onto a VM.  Not sure this is one of those situations.
 
But when you're looking at this many separate little systems it seems, to me anyway, that reliability and consistency would be better served with it standing independent of anything else.  At least then you're not at the mercy of the unexpected 'gotchas' of running within a VM.  Be they administrative or technological.  
 
Yup; here also previous enterprise "gig" collapsed multiple BI servers to running in VM's.  (along with a few other "boxes")
 
Here though had licensing issues relating to SAP changing methodologies of licensing when they purchased BI.  (IE: that whole number of processors utilized type stuff).
 
That and doing some multiple site WAN connected building management card access stuff tested using VM's but went to the direction of a dedicated box with an abundance of hardware devices connected via the LAN and WAN.
 
wuench said:
I run CQC in a VM and so far the only issues I have had were not CQC related.   They mostly centered around getting COM ports and USB devices to pass through successfully to the guest.  But I am running on the now deprecated VMWare Server not on an ESX host...   I also use RDP as my primary means of accessing CQC through android phones/tablets and there is no issues with delay, etc (as long as you assign enough resources).  But you can run the interface viewer on any windows machine, or use a RIVA client on IOS or android clients.
 
CQC itself is very lightweight, I originally started running it in a 512MB VM on XP.  I have since bumped that up to 2GB in order to support multiple RDP sessions.    
 
It is modular (as I think Elve was) so you can distribute the various components to run on machines near your resources,  although they may need to all be in the same subnet/broadcast domain to find each other.  Jonathan was a long time user and big contributor to CQC before founding Elve, so a lot of the concepts are similar.  The primary fundamental difference that I see is Elve leverages .NET as something more people are comfortable with, whereas CQC tries not to be dependent on underlying libraries in an effort to maintain stability.   So it's all custom windows/dialogs, macro language, etc.
 
 
Thats great too hear.
I think if licensing is not an issue then I too would prefer to use a VM to host CQC.
Backups and snapshotting are done by one of our systems administrators so if I did screw anything up it is great to know I can simply roll back the server to before I made any changes.
We also have a second ESX host in place so if there was ever a need of disaster recovery we can "fire up" our VM's on this host with little or no down time.
I do appreciate what people are saying Re: using a physical machine but I did run Elve on a VM with no response issues and I am of the opinion that if you have a better infrastructure in place you might as well use it.
It sounds like CQC is indeed lightweight so the fact that I can throw resources at this box is not relevant.
My initial reason for starting this thread was to see what alternatives to Elve might be out there and it seems like CQC may be the only one.
Thank you all again for your input.
 
There are other systems but it's hard to know how well they support multiple panels.  You could certainly visit the Home Automation Software link up in the resources to see which ones support the M1 and your other required peripherals and work from there.  This thread definitely took off on a tangent.
 
Back
Top