Re: Mini-ITX PC's a the future of HA (was Re: X-10 Mister House Motion sensor problems)

["Robert Green" <ROBERT_GREEN1963@xxxxxxxxx>]

"Dean Roddey" <droddey@xxxxxxxxxxxxxxxx> wrote in message
news:wGDig.145795$F_3.139448@xxxxxxxxxxxxxxxxxxxxxxxxxxxxx
> I doubt HS' choice of a mini-ITX system had anything to do with whether a
> panel is better than a PC.

Until someone from HomeSeer clues us in, we'll only be able to guess why
they did it.  But I'm not sure I'd want to look at the evidence and draw the
conclusion that they did not perform an analysis of both types of systems
(Let's agree to call it the "panel v. PC" debate) and find one route
superior.  I believe they did.  Too bad they no longer maintain a presence
here but I can't really blame them for that. Still, I'd like to hear how
their mini-ITX HA new server came into being.

HomeSeer *could* have looked at the Elks and Omnis of the world and created
a panel that handled not only security, but lots of routine automation tasks
as well.  After all, if panels are better suited to security because they
are more reliable, then wouldn't that same sort of reliability be good for
home automtion too?  HS, then, would seem to be taking an unreliable course.
But we both know that's not the case.

What's really happening is that the years of proprietary panel protocols,
inability to access networks, inability to interface with home automation
and AV gear and lots of other issues have put the mark of death on panels.
Vendor's panels (let's take Omni because I own their gear) have been
steadily "growing" in size and complexity and add-on modules to account for
these shortcomings.  I claim it's an unsustainable task in the long run
because competitors can enter the market using a mini PC as their foundation
HW and have both a cheaper and superior product as a result.  A more
powerful CPU gives you lots of options a panel can't.  Lots.

No one building a custom board can possibly compete with the horsepower and
flexibility of the modern PC, particularly as embodied in Via's mini-ITX.
That means Elk's competition can produce a PC-based alarm and HA system for
the cost of designing some interface peripherals and some software.  The
expensive custom board design and manufacture drops out of the equation.

Custom panels will take a long time to die, but they're dying.  Ironically,
Homeseer is position itself to directly compete with what it supports.  I
don't think it will be long before one security vendor or another realizes
that and pulls back on cooperating with them.  Just as security panels grew
PC-like parts, I predict HomeSeer *will* be producing other, specialized
hardwire like sensor interfaces that would remove a lot of the incentive to
buy an Elk or Omni.

For security, as Bob B. pointed out, hardwired panels still have some
advantages.  They don't depend on MS for underlying OS, for one thing.  PCs
have become incredibly robust over the last 20 years because they have been
constantly refined.  Yet because they have been running MS software, they've
taken the rap for being unreliable and needing burping and booting every
day.

> They have a PC based product.

True, but that just means they resolved the "panel v. PC" debate a long time
ago, as you did.  ADI chose an intermediate route, interestingly enough, for
its home automation design.  Much more flexible than any of the alarm panels
I'm familiar with because it's essential a micro-PC in a very small box.

> So using a panel  would have A) required them to build a panel,
> something that would cost them a lot of money, and B) probably
> make major changes to their software.

HS supports dozens of other "panels" so they're in a unique position to
design the biggest, "best-est" badass panel out there.  There's a lot of
incentive and business advantage to avoid using COTS and forcing your
clients into a proprietary design.  I believe that they're up to the task of
doing whatever they choose to do, it seems.  They chose not to emulate the
current (alarm) industry standard of custom panels.  I believe that it seems
like they are locked into PC's and that's why they are pushing the ITX but
in reality they made "panel v. PC" decision a long time ago and it's only
with the arrival of the highly reliable, fanless and very small ITX that
they could finally realize that choice in a HW product.

> So, like us, it's a choice we made implicitly a long time ago by
> creating the type of product we did.

But you did make the choice - that's all I'm trying to get at.  Both you and
HS are pioneering new technology.  Alarm panels are pretty well mapped out
by now.  You looked around, saw a platform that you *knew* could do what you
want to do and more, and went with it.

> Even if we thought a proprietary panel was superior,

But obviously you didn't think that, did you?  No.  You made a choice that
PC's were now powerful enough to support your ideas.  When most alarm panels
were designed, that wasn't so clear a choice or it was a choice alarm
equipment makers didn't want to face.

> there's not much we could do about it short of creating a
> completely new product and hiring someone to do design
> a panel and then to have it manufacturered.

It's not impossible.  Elk and Omni and loads of other do it all the time.
With their own, dedicated panel they won't have to be at the mercy of other
manufacturers and design changes, incomplete protocol specs. OS changes and
all the other issues that plague interoperation between proprietary devices.
Panels were born in a time before AV gear was so complex and before very
people had even heard of home automation.

Year after year, more and more "modules" are being added to these
proprietary panels and they get more complex (and thus inherently less
reliable) and more expensive as a result.  They are trying to communicate
like PCs and be as "smart" as PC's but they can't get there from where they
are and never will.  PC's are developing too quickly to ever be "caught up
with" by a custom panel.  The "whole house" PC is coming, and it's going to
drive panels into the museums.  People want event logs - panels stink at
that.  People want networking and web-enabling and USB and audio and video
and lots of other things that are all there on a PC but quite a bit more
expensive in the "panel centric" world.

> Both platforms have their strengths. There are things that PC
> based systems can do that a panel will never do. OTOH, a
> panel is a hugely simpler product, and simpler (as a rule)
> means less likely to break.

I think that's an outdated assumption for two reasons.  Every time I look,
Omni's adding something to its boards, as are the others.  They *have* to in
order to keep up with the expanding universe of home automation.  That means
they're moving *away* from reliability.  PCs on the other hand are at least
at the 20th major design revision, depending on how you count.  I contend
that MS, having let HW take the black eye that its software deserved, has
actually forced some PC makers to seek the most reliable components they can
find.

PC MTBF varies so radically between MS and Unix that it becomes readily
apparent that a lot of the "unreliability" of PC's lies at the feet of
Windows, not the hardware.  For that reason at least some believe that
HomeSeer made the wrong choice in going with Windows, and not Unix.  I tend
to agree.  MS has not demonstrated that it can create a reliable operating
system suitable for "appliances" let alone critical black-box functions.

> This is not to say that PC based systems cannot be made very stable,
because they can.

OK, strike that last paragraph.  You're aware of the situation.  You have to
be, of course, because your product is only as good as the hardware it runs
on.  Would I be correct in assuming that you contend with far more
software/firmware issues than underlying PC hard faults.

> But it can only be done emperically by finding a set of
> components that fails to fail, because we can't look inside
>the box.

That's exactly what I was discussing somewhat earlier.  And it's why I am so
enamored of the mini-ITX.  Via's designed it and they've been building this
stuff for years.  Lots of people are using them as servers and reporting
phenomenally long "up" times.  They have, from what my hands *and* my eyes
tell me, have come up with a machine that "fails to fail."  As a chip maker,
they're in a unique position of knowing how to design every inch of a
motherboard.

Intel dropped the ball on CPUs for a while.  They didn't realize that most
people didn't want their laps burned and the tiny fans in their PC's to fill
with dust and fail.  But that's what Intel delivered, fixated on speed,
speed and more speed.  Via ate its lunch by delivering what customers really
wanted.  Reliability, low purchase price and low operating costs.  Hold a
Via mini-ITX motherboard in your hands and you'll know you are holding a
technological miracle.

I've been building PC's since 1984 so I know what's full length-cards have
been reduced to postage stamp IC's.  They've used every clever trick in the
book to reduce board size and power consumption and increased reliability by
ditching the tiny fans.  There are "heavy hitters" in this world that change
the way things are done.  Jeff Volp's XTB is one of them.  The VIA mini-ITX
is another.

> And this is not to say that  panels don't have their problems either.
> They do have software in them and it can be wrong sometimes, and
>  the market forces them over time to get more and more complex.

I believe that it's this "forced march" to complexity that's going to be
their undoing.  You build things, Dean. You know that you can only hang X
number of additions on a framework that wasn't really designed to support
those additions before the whole thing collapses.

I also believe that as these "panels" take on more and more complexity and
continue to grow, the PC industry has stabilized and is now very, very
mature.  The hardware changes that used to occur monthly are now mostly over
with.  As time passes, and panels become more complex, the PCs are actually
getting simpler, smaller and more reliable.  I believe that the Via mini-ITX
is the perfect example of that trend.  Because they are used by millions of
people flaws are very likely to be quickly exposed.  Just looking at Via's
revision list tells you how hard they work to run down problems reported by
end users.

> We chose a PC based platform for a number of reasons. One, I
> had ten years of general purpose PC based software architecture
> to build on. Two, there were already plenty of panels, so what
> would have been the point to get into that market. Three, I really
> do think that PCs are the future of automation,  though in a more
> 'robustified' form.

Then we really do agree.  The fanless, diskless mini-ITX's are more than
powerful enough to run a house's alarm and HA functions.  Losing the fan
meant losing the one moving part that's responsible for at least half the PC
repairs I encounter.  Lose the disk, and you've eliminated the other half.
That, IMHO, is moving to just that "robustified" form.

Unfortunately, I think robustification process almost certainly means a
Unix-based platform.  People have accepted that Windows crashes because they
keep buying it anyway.  There's not much incentive for MS to improve the
product if people don't care about having to reboot.

> Four, the PC market is an enormous R&D machine that dwarfs
> the automation market, spending more every year on R&D
> than Crestron and AMX's total worth I'm sure. No proprietary
> hardware will be able to remotely keep up with that.

Exactly, precisely, indubitably!  It's probably the number one reason that
panels are doomed.  There are so many companies working together to produce
the modern PC that there *is* no way any single panel maker can hope to
approach the level of R&D.  Modern motherboards are so well-designed that
you can short the board to ground and it will not burn up the way it might
have 10 years ago.

> Five, given the above, software becomes what is important, not
> hardware, and that's what we are good at.

Software IS the most important part now.  That's why having a common
hardware platform is so valuable.  It's also why every Tom, Dick and Harry
proprietary panel and their often cryptic programming "languages" are doomed
to extinction.  Yes, there will be a lot of them around for a long time
because there's a huge installed base to contend, both of customers and of
trained service personnel.  But for the reasons you've very clearly
elucidated, the panels are dinosaurs.

> Six, the home is headed towards a network backbone,
> and those who can provide an automation system that
> leverages that network can get in for a lower cost because
> the home owner has alreayd bought much of the infrastructure
> that doesn't get 'charged' against the automation system.

Even more reasons that the custom panel's going to die.  How many times have
I read posts here about people not wanting to install another set of sensors
to feed their HA setup but they can't figure out how to make their panel
alert their HA system that an alarm has occurred?  As you point out, if your
device is already on the home network, you're "half-way home" already

> Seven, there is a tremendous range of hardware out there
> from tiny systems to multi-CPU mondo-servers, on which
> a software product like ours can run without change, which
> provides us with the abilty to scale up and down pretty far
> with a single product (which greatly reduces complexity of
> product development.)

Want to upgrade your OmniLT?  Buy the $1000 Omni Pro II model and scrap your
old panel.  That's a bad upgrade path, at least IMHO.  In the PC SW based
model, you'd buy an upgrade to your software and maybe a few extra I/O
modules.

--
Bobby G.