I built this to hide my stuff. There are 2 satellite boxes, a ReplayTV a PC 2 UPS's and a CD carousel behind the stone panels on both sides of the fireplace. The drawer units on the right were not installed in this picture and there was still some wood trim out that was not done,.
formatting link
You can't run the latest Flash on a 1.5gz single core machine.
Then you are not streaming. They all either want HTML5, Flash or Silverlight.
I only use MP3 for audio, usually 320b.
I am trying to decide on a format for video but my player handles just about everything I throw at it.
- water controller (metering, filtration/conditioning, irrigation)
- utilities (monitor gas and electric)
- weather monitoring (on the roof)
- swamp cooler (*in* the swamp cooler)
- 3 "user consoles" (my "accommodation" for SWMBO's aversion to "all this tech"
- water heater (leaks, temperature control, cycling, hooks for solar)
- garage door interface
- 2 WiFi access points
- 5 localizer beacons (to track where the occupants are currently located)
- 24 "general purpose" network drops (each independantly firewalled) etc. Aside from the "user consoles", you'd not know any of this stuff was "in" (literally, HIDDEN IN) the house. If you peek at the bottom shelf in the kitchen cupboard, you'd suspect *something* unusual was happening, here! (where the switch, UPS, database, etc. hide)
You don't run it on a 1.5GHz machine! You convert it to a form that a slower machine can process. It doesn't take much horsepower to display full-motion, live video!
Anything from a "live stream" gets transcoded (if we want to view it "live") or "converted and stored" if we want to view it at a later date.
E.g., I deal with audio in all sorts of formats -- but my network speakers can only *play* it in my proprietary format. So, anything "live" gets transcoded ("converted on the fly") to that format. Anything that I want to replay, later (e.g., CD rips) gets *converted* and stored (it should be obvious that the "converter" is the same device that does the transcoding; the difference being whether the results are stored or consumed "live")
I keep a 500G archive of MP3's (for PMP's and "PC usage"). But, this is "distilled" from my lossless format, elsewhere.
I have to choose a format that is appropriate to supporting multiple concurrent clients "tuned" to the same media stream. I don't want to have to double the capacity/horsepower of my server just because I have doubled the number of clients listening to a particular radio or TV program (at the same time). E.g., if one client wants to listen to an audio program, the load on the server is exactly the same as if
30 clients want to listen to that same program.
Doing the same for video makes sense (imagine you were delivering video to displays mounted on the backs of airplane seats; or to 100 rooms in a hotel; or to dozens or hundreds of students in a school; or hundreds of seniors in an assisted care facility; etc.). As each of those clients can potentially "drop" packets, you don't want the clients all trying to re-request those dropped packets from *the* server! That means more clients results in more workload for the server.
Instead, you want a format that allows clients to request dropped packets from *peers* (similar to how a torrent works). The more clients you have, the more *peers* that you have to get assistance from! So, you can keep adding clients without having to move to bigger and faster servers.
A solution that SCALES like this has to carefully consider every aspect of its implementation -- so you *can* shift some of the responsibility to peers without forcing them to take on an extraordinary amount of work.
We currently have one SFF PC in the living room that acts as a DVD player (we check out lots of DVD's from the local library -- probably
10-15 per week). It also has SWMBO's daily exercise videos stored on it. And, can act as a PVR with the HDHomeRun boxes.
The other TV's pull their "content" from the HDHomeRun *live*... or, from material stored on that little PC delivered locally via these Optiplex FX160 boxes.
So, if we want to watch "video" in any other room, we have to turn that PC on, first.
In the future, the STB ("network display" for want of a better word) for a particular TV (display) will be powered up by the PoE switch. The software that makes it a "network display" will be delivered over the network. And, the "content" of interest similarly delivered (if that means powering up an HDHomeRun for "live TV" then the network connection to that HDHomeRun will be powered up thereby powering up the HDHomeRun, in the process).
None of this requiring us to push buttons on a "front panel" or a "remote control" (which you would then have to keep track of!)
The "real" PC (i.e., it actually has a disk in it) is located in the living room. The other TVs are located in other rooms. None of the TV's are smart.
These devices are on their own private network -- so they don't muck with any of my work machines (*or* talk to the outside world!).
The "real" PC has to be up and running in order to make available the software image that the little Optiplex FX160's download (TFTP/NFS) in order to display video on their respective TV's.
If I had another box running 24/7 that could serve up these images, then the FX160's could fetch their software and then send a "magic packet" to the "real" PC to gain access to it's disk drive (movies).
But, the only box that is up 24/7 is on my "work" network -- not accessible by these machines.
[I could install a laptop drive in each of the FX160'S (or even thumb drives with the required software configured to boot via usb) but, that's not how I want these devices to operate in the final implementation (so, I want experience to see how things can fail
*if* they fail)]
I would think a centralized technology closet, basement, or spare room would make more sense with everything home run back to that tech core. If the amp fails, you replace the amp, not the speaker and network connection. Decide to change things, just swap a few cables in the tech closet.
Yep, same here.
All of my audio is 320K MP3 as that format is widely supported by most devices, including my smart phone and the stereos in our cars.
Video is generally h.264 MP4 with AC3 audio. I use 30Mbps bitrates with my personal home videos, and 9Mbps for anything that I can easily replace (TV shows, movies, etc.).
I don't really have much to hide in the living room. My tiny M6500A media streamer sits nicely next to my Sony Blu-Ray player (which also streams Netflix). My audio receiver sits on the shelf below.
The majority of my technology is located in my computer in my home office. It records and streams my shows, as well as control various lights, pumps, and fans around the house.
Only if you're streaming from a web server.
Local streaming uses different protocols over your home network.
Me too.
Until recently, I used 30Mbps MPEG2 for all of my videos as it had the best quality, was widely supported by most devices and video editors, and was easy to stream.
However, h.264 video has now reached a similar compatibility level and offers better quality. So I now use 30Mbps h.264 video and AC3 audio in an MP4 container.
It runs 365/24/7 anyway as it also controls our outdoor lighting, our hot water recirculating pump, and our fresh air ventilation fan.
It also runs backups every night while I am sleeping, in addition to the hourly backups it performs during the day.
In the middle of the night it also processes the TV shows I have recorded, removing commercials, renaming the files, and moving them to my desired directory.
I have a UPS (Uninteruptable Power Supply) for my computer, so it all keeps running even when the power goes out.
Those network cables have to come from somewhere right? Maybe a network switch or router? That's not much different than my centralized PC setup.
Hmm.. I have food in my kitchen cupboards, but whatever floats your boat. :)
72 ports? Wow. I've got a five port router in the office (2 printers, the computer, a VOIP phone, and a line running to an 8 port switch in our crawlspace).
The 8-port switch runs to my media streamer, my blu-ray player, my daughters computer, a wireless access point in the attic, and the rest go to wall plates that I never use anymore.
I've used other options in the past, dedicated timers and whatnot. I much prefer the centralized location as it's easier to make changes and upgrade when needed.
My phone system records calls too, but if it is unavailable due to a power or network outage, my VOIP service records the call and sends the WAV file to me by email too.
My computer is always on, so I make a strong effort to select components that use little power and are very quiet. Once I turn the monitor off at night, my computer typically uses less than 90 watts. I use quiet fans that are virtually silent so there is no noise overnight either.
I use Insteon switches to control lights, a pump, and a fan. Then use an Insteon interface and software to control them with my computer.
This allows me to control the lights on our detached garage, even though there are no separate cables running to that building (it's out of wifi range too).
Correct, my house is all I'm concerned with. I'm not running a hotel or business complex.
That puts a lot of kit in a place that then needs to be aggressively cooled. And, takes up a lot of space! Think about how many "field" leads are involved there -- all the speakers, the microphone on each, the drivers for the irrigation solenoids, video output for the TV's, wires from all the alarm sensors, inputs from the weather instruments on the roof, antennae/CATV feeds, phone lines, coax from all the cameras, control signals for the PTZ camera bases, etc.
[I've already got ~3000 ft of CAT5 in place; I'd need more than that if I was running field wiring to the "equipment cupboard"]
And, how many PC's do I stuff in that closet with countless "I/O cards" to interface to all of this field wiring?
It's much simpler -- and more economical -- to put the computer *at* the I/O. This also lets you *add* computers at other sites just by snaking a length of CAT5 to the location -- assuming you don't already have a "drop" there to exploit. E.g., there are 4 uncommitted drops in the living room so I can put a floor standing "network speaker" in each corner of the room without having to run "speaker wires" from some central closet. Likewise, I can add a "network display" (aka TV) in any room just by connecting to one of the uncommitted network drops *in* that room.
My original implementation had 11 PC's in an "equipment closet" talking to dumber processors distributed around the house. This would have cost us the walk-in pantry (SWMBO wasn't keen on that idea -- in a house that is already short on storage space).
Increasing the computational horsepower in each of these distributed nodes allowed me to get rid of all of the PC's save one (COTS RDBMS). And, lets the system grow by naturally adding more "motes" as the number of I/O's increases (add another camera and you don't need to add another PC to do the real-time motion detection *in* that PC)
Only one box here runs 24/7 -- the DNS/TFTP/font/NFS/NTP/RDBMS/etc. server. It's implemented with one of those FX160's as it sits under my dresser (no display or keyboard; telnet/ssh to it when I need to do maintenance)
My goal is to only run what I need to run based on *when* I need to run it. E.g., the answering machine only powers up when the VOIP gateway indicates an incoming call; the irrigation system only powers up when something needs to be watered; the network speakers are only powered on when I need to listen to something, etc.
I have a UPS on each workstation (I think 11, at last count). The switch for the automation system has a pair of SmartUPS 1500's. But, I will replace them with a 48V unit (lets me get more efficiency in the PoE capability as I don't have to develop the 48V just for the switch; let the UPS develop it for the battery pack!)
The alarm system has it's own battery backup (literally a battery as it is PoE powered, as well).
There's no way around needing a switch (wireless is too vulnerable and means you then need to distribute power via some other means -- and back it up in all those locations!). I you put a bunch of PC's in a central closet, then you need a switch to let them talk to each other, etc.
Powering everything from the switch lets me avoid lots of wall warts around the house. And, lets me back up (power) everything from one place. As the switch lets me power down individual "drops", I can scale back the services that I offer during a power outage (i.e., probably not important to water the citrus trees if they *should* be watered but power is down).
We have *no* storage inside the house -- beyond the bedroom closets, a "guest closet" (that also stores the vacuum and carpet cleaners) and the kitchen pantry and cupboard.
The master bedroom's closets are full of clothing and some artwork (can't store art in the garage!). The office closet is full of test equipment, "supplies", magnetic media, etc. The guest bedroom closet has laptops, A/V equipment (stored), laser/dvd's, etc.
I've commandeered the lowest shelf in the pantry (I never to know whether to call it a pantry or cupboard; it's the larger of these two "kitchen storage spaces") plus the floor beneath that shelf. The UPS's sit on the floor. The VOIP gateway and HDHomeRun boxes connect to antenna and phone feeds that terminate/originate there. All of the network cables terminate on a patch panel -- from which they can be cabled to the network switch that sits there. An SFF PC sits on the shelf with external USB drives tethered to it (to implement the data store for the RDBMS that runs on the PC). There's a 1U console on slides that I can pull out to "talk to" the PC -- and, from there, to the rest of the system: It's hard to imagine a *smaller*/denser implementation!
24 of the ports are "uncommitted". E.g., there are 4 drops in the living room, two in the kitchen (counters), 1 each on the back and front porches, 2 in each of the three bedrooms, 2 in the dining room, 3 in the family room, one in the front hallway (and I'm forgetting a few).
The remaining 48 are wired to bits of technology hidden in the ceilings, walls, etc.:
- HVAC & swamp cooler controllers
- water, utilities (gas/electric) controllers
- 11 security cameras
- 10 network speakers (ceiling or high on walls)
- laundry, garage door, water heater, landscape lighting
- 5 localizer beacons
- 2 wireless access points
- 2 UPS's
- CATV, POTS and DTV interfaces etc.
The drop in the office connects to the (regular) 24 port switch that feeds the machines located there (2 PC workstations, 2 Sun workstations,
2 X-Windows terminals, 2 1U servers, 2 2U servers, SWMBO's computer and laptop, 3 printers, several COTS NAS's, a couple of pieces of test equipment, etc.).
The drop in the guest bedroom talks to the (regular) 16 port switch that handles the multimedia workstation, my DNS/TFTP/etc. box, another printer, a (docked) tablet PC, several FX160's that implement my "custom" NAS, etc.
[I have a LOT of kit! Perhaps you can understand why I am obsessed about hiding any *other* kit that the *house* might "require"?? :< ]
Ah, but the distributed system *appears* as a single cohesive system! It just looks like dozens of "cores". Instead of having all of the field wiring coming to one GIANT multicore machine, the cores are located with the I/O's.
[The "program" that runs the irrigation system does not NEED to run on the node that has the irrigation system I/O's wired to it! The system can dispatch that program to some other node as it deems appropriate (because the irrigation controller might be busy detecting commercials in some video that is being recorded by another processor, somewhere). Likewise, if the system needs more computational resources, it can power up a node that is currently "off" and use it's CPU/memory to address some other task for which it needs resources. When done, it can power that node down to conserve power.]
We don't use a VOIP service. We have POTS service *to* the house. We use VOIP *inside* the house. I.e., a daemon watches the VOIP gateway for signs of an incoming call. When detected, a task is dispatched (fetched from the RDBMS's persistent store) on a "processor with available resources" that answers the call, decides if it is someone that we want to talk to, decides if we are "available" (i.e., not asleep, in the shower, in the back yard, etc.) and then notifies us of the call -- which we can elect to route to the answering machine, etc.
If, for example, *I* am calling, I might want to issue commands to the house ("water the roses", "let me know if I left the garage door open", "give me my messages", "record a message for Bob when he calls, later", etc.). Likewise, if we're out of town and have a neighbor watching the house, they can call and tell the house to perform certain actions (instead of having to come over, let themselves in and tell the house directly).
When walking around the neighborhood, I can carry a cordless phone (NOT a cell phone!) and talk to the house -- or, have the house talk to me ("A package was just delivered for you", "Bob is waiting for you at the front door", etc.)
My machines have lots of I/O's -- tablets, cameras, motion controllers, SCSI peripherals, etc. So, lots of I/O cards in each machine. I run three monitors on each workstation so multiple video cards. etc.
[One reason that I have so many different workstations is due to the number of I/O's that I support; it's just not possible to connect everything to *one* machine]
If I leave a workstation on 24/7, it shows up in our electric bill!
I'll *buy* a "solution" for lighting. It's just too much work to try to control loads with "custom" hardware. (And, I don't want to be in that market)
I've invested a lot of effort in the system design and the hardware/software implementation. It would be foolish for me to be limited by "a family of four".
As I want to show folks how you can make products accessible, I want to be able to demonstrate that I'm addressing *big* projects, not just "token" projects.
I can easily see taking modules from my design and deploying them in an assisted living facility ("help I've fallen", video conferencing so residents can "visit" with their neighbors while otherwise not mobile, local HVAC controls, etc.)
Or, in a business office environment.
Or, in an "institutional" setting (school for blind).
Or, in a hotel.
etc.
Individual modules can stand alone to address particular needs (security system, IP cameras, HVAC controller, irrigation controller, answering machine, etc.). By designing them as appliances (instead of "software running on a PC"), they can be "ported" as-is... no need to redesign "PC hardware and software" to fit in an appliance form factor!
[I have never liked "do overs" in product development. Get it right the first time, then move on to some other challenge. "Version 2" is nothing more than a chore, in my book!]
And to think this thread started because I wanted to record NCIS for my wife ! Y'all are way way more involved in computerizing/automating things than I have any desire to . To quote Don Williams , "I'm just a country boy."
The "involvement" is a consequence of the immaturity of computerized products. Even the "appliances" aren't (yet) "turn-key".
Consider how "involved" it was to cook *inside* a home, wash clothes there, refrigerate the interior air, etc. before we had appliances to do those things. Consider how "involved" it was to operate an automobile. etc.
But, those technologies have all "grown up" and become mainstream. (people don't even look under the hood of their vehicles anymore!)
I just have the one computer running everything, so it's the only box that runs 24/7 also.
11? Wow, that sounds expensive, and space hungry.
I have a Cyberpower CP1500PFCLCD for my computer, monitor, cable modem, and router.
My network switch is not on a UPS, since all the stuff that accesses it goes out when the power goes off anyway.
My UPS isn't running very long these days before it shuts down (about 10 minutes max). I replaced the batteries last year (cheap knockoffs) but it didn't help. I just ordered new genuine Cyberpower batteries to give it another try. If it doesn't help I plan to buy a larger UPS and delegate this UPS to lesser importance devices.
That switch must have a large power supply in order to power all 72 ports?
With that many devices, I can see why your network powered system would be a smart idea.
We dropped the standard telephone service years ago. They were charging over $100 a month, provided no caller ID or other services, and only local calling to our little town. All of our family and businesses we dealt with were long distance.
Once we got cable internet, I switched to VOIP for about $20 a month, full services, voice mail, and free long distance anywhere in the country.
Kudo's to you for thinking big. I don't think most people need that kind of thing in a typical home environment. As it is, my wife thinks I've gone overboard with my simple setup. :)
There'd be no practical way of me getting all the field wiring *into* one computer!
So, the little 24/7/365 box just provides the "core services" that all of the other boxes need when they come online. The same approach is implemented by the RDBMS server in my automation system (it's the only place where things can be "remembered")
Most of my machines are too big to be able to put them on a shared UPS (and be able to power more than one up at a time). The UPS's also act as "outlet multipliers" and "common power switches" -- I'll plug a workstation and its attached peripherals into *one* UPS; the next workstation (and *its* peripherals) into the next UPS, etc. So, a workstation appears to have a single power cord instead of many.
E.g., the workstation on which I prepare my documentation has the PC, a film scanner, an A-size scanner and a B-size scanner attached to it along with a DLT. Each, of course, has a power cord. When I'm done with that workstation, I can push the power button on the UPS and ensure that all are powered off. In the event of an electrical storm, I can unplug the UPS and know all of those devices are isolated from the mains.
Repeat this for the CAD workstation, multimedia workstation, each of the two Sun workstations, SWMBO's computer, the 1U servers share a UPS (because they have few I/O's and are never powered up together) and the 2U servers share another (same logic). This email/WWW machine has its own UPS as does that little DNS box.
As my monitors are "shared" between workstations (I connect each monitor to two workstations and then use the A/B switches on the monitor front panels to select which workstation's output I am viewing), they can't be plugged into a "workstation-specific" UPS (cuz the UPS for that particular workstation might be powered off while the workstation that is supplying video is powered on!). So, there's another UPS that just keeps the monitors "up".
Another does the same for the NAS boxes (again, they aren't associated with any *particular* workstation so can't logically be tied to a particular UPS)
The expense comes in the form of the batteries. :< Buying them in lots of 20 helps with the price. But, it's not the sort of purchase that I enjoy making!
For the most part, I just need the UPS's to bridge short brownouts and blinkouts. Actual, prolonged outages are very rare, here. (OTOH, a single "blink" can screw up the rendering of a multimedia presentation or the layout of a printed circuit board; the price of a battery is a pittance compared to the time lost!)
Ditto for the office. My only risk, there, is a file transfer to/from a NAS, FTP service, etc. And, as those things are under my control, I can always restart them at a later time.
The automation system, however, is a different story. I can't afford to lose telephone, HVAC controls, security, etc. just because the power glitched. And, having everything reboot can quickly lead to inconvenience (though I can boot everything in less time than a PC takes to get up and running -- the RDBMS being the slowpoke in the lot)
The PoE switch supplies 48V to the devices on the ends of the network drops. So, having a 48V *battery* in the UPS (instead of the 24V batteries in my current UPS's) seems like it should be more effective at keeping things "up" when the power is out.
Yes. Worst case, about 1100W (in addition to what the switch uses itself). But, that assumes each port is drawing the maximum 15W that I can deliver over the network. I don't use that sort of power on every port (imagine a dozen bookshelf speakers consuming 15W?? A dozen IP cameras consuming
15W each??).
It's the only way all of that distributed kit can be practical! if each device had its own "local" power supply, you'd always have to wonder if a (remote) power supply had failed and that was the reason why you couldn't "talk" to a remote node. With control of the power 8at* the switch, I can know that power *is* being delivered (diagnostics in the switch) so if the remote node isn't responding then it's cable has been cut *or* it is fried. I.e., you can skip the "is the device plugged in?" part of the troubleshooting.
Imagine in an office or "institution"... someone unplugging a wall wart and crippling a device! (Consider the implications if that device was providing a security function!)
We have no long distance needs. We used to buy $20 "calling cards" (2c/minute) but found that we couldn't use them up in the ~year allowed.
I think our local phone is ~$30/month and SWMBO's cell phone is another $8-10. For the few times when she wants to phone her sister, she'll use the surplus minutes on her cell phone -- to help burn off the "balance".
Cable, here, has a bad reputation for service and availability. Most folks drop the service and switch to Dish for TV and DSL for ISP. You can see the number of homes that have had wire problems as the cable company just lays a cable ON the soil and promises to send someone around to bury it, "soon". Their idea of "soon" apparently differs from most folks'! (i.e., years!)
I spent most of my career addressing *markets*. Now, I address *needs*.
My goal is to demonstrate how you can "design for accessibility" as the number of "deficiencies" (avoiding the term "disabilities") that are apparent in the population suggests that's a common problem (7% of men are color blind; 10% of men over 50 develop essential tremor; diabetes and macular degeneration costs folks their vision; age costs folks their mobility; Parkinson's; ALS; etc.)
But, a "white paper" approach would just be received as "gee, that's interesting" with no real consequential followup. A "token" application trivializes the effort involved and the impact (design an alarm clock that can be used by people with any/all of these "deficiencies").
So, I needed a substantial project to illustrate different challenges (to the user interface) and how they could be consistently addressed. And, something that is "exciting"/interesting (no one cares about an "accessible clock"!). Put someone *in* a "device" with which they can interact in a variety of ways and most folks want to poke at it to see what it can do, what its limitations are, how it addresses particular situations, etc.
It becomes a MEMORABLE experience. So, hopefully, they take that memory with them as they begin to address *their* designs. And, maybe, think about the assumptions that they implicitly "encode" in those products.
HomeOwnersHub website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.