That’s not a standalone sentence. This was the paragraph regarding how I was using the term service failure.
If a network would improve QOS with twice as many servers, that to me is a service failure, not a technology failure. The devices work fine, we’re just overloading them.
If adding more hardware solves the QOS, then that’s my definition of “service failure.” In that case the devices are OK (the case where adding more hardware improves QOS).
My very first sentence in that post said there might be something weird going on, but based on my many discussions with support, they identified most of the issues I was experiencing as “hiccups.” (Their word.)
Do I know for sure these hiccups would be reduced with more servers? Nope, and never said I did. But plenty of companies run similar or more complex services in the US with a higher degree of reliability. Whether it’s cloud based video services, alarm.com, mint.com, utility companies, Google maps, Dragon Nuance, etc, “hiccups” are rare and costs to users are often low.
Remember the initial problem I mentioned: a Hello Home Action set for 8 a.m. that does one thing, change mode from Asleep to Home, failed to fire after running successfully for several months. And support showed no concern about the failure, according to them the servers “hiccuped” and I should just reset the timer manually.
If that wasn’t a musical chairs issue caused by an overloaded system then we have a home automation system that can’t reliably run a daily schedule to set one value.
I could indeed be wrong but my personal belief is that if SmartThings couldn’t handle a simple time based schedule for other reasons then nothing would work. And it would be failing a lot more often for a lot more people.
So my guess is that it’s a scaling issue. Growing pains. But as I’ve said from the beginning, I could indeed be wrong in that guess.
It doesn’t matter, though: whether it’s poor design or bad technology or QOS due to overload, the end result for me is the same. I can’t rely on it.
I use many Internet-based services. Some are by subscription. Some are free up to a certain volume. Some are totally free. The business model doesn’t change my reliability requirements.
Other people will have different requirements, and different means of dealing with reliability failures.
But going back to the CTO’s comments, if my lights don’t go on in the morning, SmartThings is always the first thing I suspect. Not the power company, the GE light bulb, the fixture, the Philips bridge, or my own logic. And over 8 months, I have yet to be surprised after further analysis. SmartThings it always was.
We’ll see if things are more reliable with ST out of the picture.