Slashdot Mirror
Embedded Linux at COMDEX
discovercomics sent us a nifty report from MSNBC telling the tale of the Embedded Linux Devices that were present at COMDEX. They talk about the IPAQ, The Yopy, the Axis 2100 network camera (which is cool, but expensive), and more.
-----
Klactovedestene!
I've seen a number of embedded projects hit serious and expensive delays because of bugs that only manifested deep in a kernel for which no source was available (or was not available at a cost a small company could afford). This has two effects that will promote Linux:
1. Proprietary real time OSs can add substantially to project cost and time. Further, this is hard to predict. This is especially true if the project is not exactly what the OS vendor initially had in mind in terms of target hardware and application.
2. Engineers HATE long unpredictable delays in their projects caused by lack of access to stuff they think they could fix if it was available. This is driving a lot of internal motivation to try Linux for embedded projects whereever possible.
Of course the Linux hype is a factor. For example, it may well be one of the reasons that some projects look at Linux instead of eCos, which is also open source.
I respectfully disagree with you. I think the reason Linux is and will remain to be popular for embedded applications is that each vendor doesn't have to recreate the wheel. In this case, Axis didn't have to write an operating system for their camera, or a http or ftp server either. Their little web server is some open source web server (no, not Apache). So rather than create an OS and a web server, they just pull together what they need from the open source community. After all, isn't interchangeable parts what made the whole industrial revolution thing work in the first place?
I'd like to make a counter point as to why I think Linux would make a good choice for embedded solutions. Note that I have several years experience as an embedded software engineer so my opinion my be a bit biased :)
Let's start with product development. When it comes to the operating system, whether it be a simple task switching executive or a full fledged operating system like Linux, WinCE, QNX, etc., there are 2 routes to take. One, roll your own or two, get one off the shelf. Rolling your own is extremely expensive to a company. Not only does it cost money to pay for the engineering, but it takes engineers away from working on the actual application of the product. That leaves getting an off the shelf opoerating system. (Admittedly, off the shelf operation systems won't work in all cases, but generally they get job done.)
To continue with product development... The ability to run an embedded product's code on your development workstation and the embedded device is a tremendous advantage. Host side development tools tend to be easier and more reliable. The API on a Linux workstation happens to be identical (with vfew exceptions)to the API on an embedded device running Linux. WinCE has a similar but slightly incompatible API to Win32. Target side development tools are more difficult and less reliable because they tend to rely on precisely setup configurations. Not to mention that there is an added step to trying out new code when you fix a bug. With target side debugging you have download the new code after each build which could take longer than the build depending on method of communication with the target.
Most readers of Slashdot know about the Linux kernel's configurability and modularity, so I wont spend too much time on that. Suffice it to say you only compile in support for stuff you need like SMP (which very few embedded projects need).
Networking. More and more embedded devices need networking support. The Linux kernel with various userland programs provides a rather complete set of networking capabilities. Whether it be serving web pages describing system status, web browsing Slashdot on a train through a web pad, or some mundane SNMP control, Linux will fit the buill.
As to Linux's "bloat", when Linux first appeared on the scene almost 10 years ago, 16Meg of RAM was prohibitively expensive for mony desktops, much less embedded devices. However, 16Meg today is nothing. 16Meg will fit on a single chip instead of several SIMMs. The capacity of hardware has grown much faster than the size of the Linux kernel.
You have 3.1415926 gallons of gas. Kewl!
It is 1.78324 miles from home to the Post Office
You have new mail.
The car is emitting 32.1 ppm SO3
It has been 2999.98 miles since last oil change
Radar detected, Cloaking Device enabled.
The Right passenger weights 183.772 pounds
Tire air pressure (PSI): 36.3 LF, 34.9 RF, 28.3 LR, 37.0 RR
Drag coefficient 0.28 @ 65 mph
Velocity 57.42 mph, fuel usage 36.11 mpg.
Scanned local radio stations and found 3 Country, 6 Hispanic, 18 Lite Rock, 2 Hard Rock and 5 Pirate.
Audio Storage: 5,772 MP3 recordings.
Um. This would rule.
--
A feeling of having made the same mistake before: Deja Foobar
Modded down because I acknowledge there are some rather nice non-Linux solutions?
Pussy.
BeIA
Real Audio, MP3, SSL, CCS, PersonalJava, Javascript, email client and much more in 6 meg.
QNX
Just a damn fast browser. Alebit lacking a few features.
The Linux Internet Appliances seem to take longer to bring to market than the other solutions, at least that is the case with QuBit and a few others. The Linux version will follow the BeIA version by some nine months, I hear.
Information Appliance Comparison by Be, Inc.
Like Gortician gives a fuck.
Get my free Hitchhiker's Guide Tribute Novella:
I had a Palm III just up and die on me one day. No combination of the buttons would bring it back to life, and I treated it with the utmost of care. Fortunatly Palm send me a replacement unit in 1 day and I've had no problems since.
Good thing HotSyncing is basically backing up your Palm every time you run it (execpt for the address book, those records never seem to make it back on the palm, Grrr). I'd bet the Palm Pilot users are among the best when it comes to backups for just this reason (even if these same people have never backed up their home/work computer).
I read the internet for the articles.
The current wave of devices with embedded Linux as their operating system is going to be a relatively short-lived one IMHO. Not because Linux is particularly bad for this sort of thing, more because Linux isn't particularly good for this sort of thing.
The fact is that no matter how hard you try to cut Linux down it's still too bloated for a device that doesn't require the full functionality that a modern OS provides. What need does my camera have for pre-emptive multitasking or virtual memory? When is my wristwatch going to need a stable threading model or SMP support?
The answer to these questions is of course, they don't. A small, dedicated operating system is easily able to handle controlling these devices, and in a much more compact and efficient manner. Just as Linux is having trouble scaling up to big iron, it has trouble scaling down to small devices.
The real reason Linux is being touted as the solution to hardware manufacturer's problems is simply that it is the flavour of the month in the eyes of the industry. As a succesful IT consultant I'm getting loads of interest in Linux from companies looking to cash in on the tech-savvy image that comes with Linux, rather than focus on the benefits it can bring in certain cases.
Remember, last year it was all Windows CE, this year it's all Linux. Neither was the solution, and neither will last. You can't put a bulldozer to work building sandcastles and expect to get perfect crenellations.
Jon Erikson, IT guru
It looks like the manufacturers are split about 50-50 between the people that are using Linux because they should and they believe in free software, and the rest are looking to shave a few more %% profit by cutting out wince. What are the odds you'll ever see the kernel mods done for the Acer PDA? Compare that with Axis, who has ported Linux to their own architecture and have the whole pile available for download on their site.
Will we be seeing any GPL compliance suits from all this Linux use?
I'm using one of the Axis 2100 Cameras for security and it really works great. It runs 24/7. I capture the images on a Linux box and filter them out so I'm left with only images with "motion" in them. Works great. Everything's been rock solid for the last month, since the last change I made to my motion detection software. Now I just need a bigger hard drive for all those images...
That's a rather scary idea there. So some guy with a gun gets only 30 seconds up the street from you... I think I rather not have him return my way after the car stalls. LoJack and a cell phone to call the police is what I'd rather have.
III.IIVIVIXIIVIVIIIVVIIIIXVIIIXIIIIIIIIVIIIIVVIII
Of more interest is the 2120 (about 850UKP), which has built-in motion detection, 25fps (PAL) and it's weatherproof (I think. I read it somewhere but now I can't see it in the spec). Check out the live feed of 45 and 5th and stress test one of these beauties.
To all the people who keep asking why anyone would
want Linux on a PDA, here's a few ideas, and why
as soon as my iPAQ shows up, Linux is getting installed:
1. Python can be installed, this would be one of
the most incredible time savers for me, as it would keep me from having to go back to my desk just to whip out scripts.
2. Think, 802.11, and a real telnet client.
3. Development. You _can_ develop for these devices with just a copy of gcc, forget WinCE where every time I see anything about coding for it, it's right along side a long description of all the expensive costs involved.
4. One of my personal pet peeves with WinCE, it's impossible to change the font size as far as I can tell, if you can I certainly haven't found a way.
5. Palm's screens are too small, as that whole lower third is taken up, so even with the IIIc (must have colour, it's not readable without it) the screen size is next to nothing.
All in all, if Apple would release an udpated Newton, I'd buy one in a heartbeat, as it didn't have any of the problems I have with current PDA type devices. Though it still wouldn't address the idea of running Python or whatnot on it, but there were self hosted Newton Development enviorments. The next best thing is certainly not PalmOS, or WinCE, and Linux at elast _I_ can make it into what I want.
The major difference between any M$ OS and any Linux distro (apart from the fact that Linux works, relentlessly :-) is that M$ insists on being in your face all the time and charging you all it can all the time.
They don't have a sustainable model. They exist on the churn. People say, "Nah, this is good enough." and M$ starves for income. Diminishing revenues stops investors, M$ cash reserves dry up and M$, not having a sustainable business, disappears.
Just like Adam Osborne did with the Osborne 1. Different reason (leaked info about the Osborne 2 dried up sales for the 1,) different time scale (M$ can bleed a lot longer,) but same result. One trick ponies die.
MSBPodcast.com The opinions expressed here are my own. If you don't like 'em... Think up your own stuff.
They didn't mention the iRobot. This thing was shown at Comdex. Its robot that is controlled via a web browser. The robot runs on linux, uses a wireless connection.. there are a bunch of specs at the site.
CNN has some coverage of the irobot here.
-gerbik
Ugh, I cringe at the sound of that stupid name ripped off from Apple's moronic nomenclature strategy. You need to hire the FBI just to track down an available one, and the battery life on it stinks (3 hours, as compared to the 8-hour Cassiopeia). And it ships with Windows CE, so you have to do a serious retooling just to get Linux on it. Considering its availability, buying an iPaq is about as sensible as buying Rambus.
"Ancillary does not mean you get to rule the world." --U.S. Circuit Judge Harry Edwards, speaking to the FCC's lawyer
I'm not sure what your experience with embedded systems is, but your views on linux's aplicability reflect that you either do not have experience with or are predjudiced towards another type (and certainly not CE as well).
For starters, the linux kernel can be as compact as you like it. RedHat and Caldera are not ideal implementations in this case. There are several distrabutions and tools available. Some are available for free (Note, I just pulled the first link off of google - I didn't see the distro we use on this list).
In development, most of our systems boot off of a compressed kernel floppy, in the field, our systems relly on DiskOnChip(DOC) and so forth to store the kernel and proprietary software.
Stability is the key thing - that and usually script management, remote access (for resetting and reconfiguring devices), security (SSL can be tiny) and so forth. The best part is, I pay for the DOC and not for the distrobution (note: some of the mini distros are for expensive, some are free).
There are more embeded systems than just consumer electronics. We do hefty amounts of utility work, percision measurement, robotics, data collectors and so forth...
Linux as an embeded OS isn't a fleeting thing, we've been doing it for a few years. I've also implemented systems using CE with varrying degrees of success (sometimes our clients demand it, sometimes its the better solution). The bottom line is that it really depends on the a couple of things: application, development cost, planned maintenance cycle, product lifetime, product downtime, support cost, and unit cost. Yes, you can cheapen a development by going with CE, but expect higher downtime, a shorter lifetime, and higher maintenance costs. I would sooner create something that I can rely in though. We have linux products 700 miles away, still happily transmitting useful signals to our customers that have not been touched since the day of instalation four or five years ago. I can't say the same things about some of our CE products of even last year.
Maybe we're one of the few companies which does things right (though I don't think so - none of us are kernel hacks and such), but I'd just say we're pretty good at designing a solid product.
If it will fit on a floppy, its small enough for an embedded solution. I don't know, maybe Linux is no good for a camera, quite possible - I've never designed a camera. There are however, systems which can and do benefit from linux. Think outside the box of PDAs, cell phones, camcorders, and geek toys.
You say you want a revolution?
A small correction but perhaps notable:
Last year the Linux Business Expo was not in the main hall as the author suggests (that would include both the north hall and south hall of the Las Vegas convention center) but in the Hilton Hotel next door. This year, we were in the more distant Sands hotel. This positioning sucked - at least from my standpoint in the LinuxFund.org booth in the ".org" area sponsored by LinuxMall - but I don't think it was meant to in anyway marginalize the Linux Co's.
I think Linux received far more floor space at the Sands than it did last year in the Hilton. Perhaps with more companies looking to be a part of LBE and more of them buying the large "pavillion" type floor space areas complete with seats for presentations and even a boxing ring, - the Hilton didn't have the floor space the LBE wanted so it was moved.
The end result, we got about half the traffic we did last year. Others I spoke to agreed. Last year many Comdex attendees did not realize that there was a deidcated Linux Expo just next door, and were excited to check it out when they learned that "Linux" (some new fangled thing the market was excited about)was there for their exploration. We were not able to draw that crowd in this year when we were a shuttle bus ride away.
Here's the upside for the LBE: - We grew while the rest of Comdex shrunk. Maybe this will allow us to command some floor space actually within the Las Vegas Convention center next year.
--Aaron Greenberg
a massive Beowulf cluster seems pretty "big" and "iron" to me.
The term "big iron" (in my experiance, at least) usually refers to very large SMP machines (in my experience, at least). They're considerably more difficult to tune an OS for than clusters, mostly because memory and communications are laid out very strangely to allow both fast and high-bandwidth access by a huge number of processors. Supporting hardware failover and hot-swapping in these systems isn't a cakewalk either.
Linux, in various forms, has started to tackle these issues, but a kernel optimized for single-processor or small-count SMP machines won't work well on a huge SMP box, and vice versa.
The point is that Linux is currently still missing a lot of features that traditional big iron operating systems already have, and probably won't implement them very well without a code fork into "small iron" and "big iron" kernels (which actually would be a good thing, IMO).
Come to think of it... Cars should be networked, that way you could hack the jerk on the cell phone in front of you and make him think he's out of gas! =)
--
A feeling of having made the same mistake before: Deja Foobar
" We're about to see an explosion of all sorts of devices that have Linux running inside. "
Someone better check out that bug! If these linux devices keep exploding, someone could lose an eye. And I wonder if it's just these devices, maybe my gateways will explode too. Oh God! Save us all!
I have lost count of the number of announcements and demonstrations with regard to Linux being squeezed into all of these weird systems.
Quite frankly, most of them have been hot air. One or two have actually managed to demo a product. None of them seem to have actually produced anything yet though!
The only reason i can think of for this is that embeded Linux systems sound great on paper, and look atractive to the accountants ("Hey, we can save money on the OS!"), but in practice the concept fails to deliver.
You only need look at the sucess of properly designed embeded OS's to see that the market leaders are companies such as Palm & QNX. These are companies who have created their embeded operating systems from scratch, for that very purpose.
Systems such as Windows CE & Emebeded Linux havn't done so well (Especially in the case of Linux), simply because they have not been thought out properly. Designers are attempted to apply a desktop solution to an embeded problem. That never works!
Linux may scale well upwords, but it doesn't scale well downwards. Thats just reality, and these designers need to get over that. It seems embeded Linux so far is a total no go.
T. Lee
Embedded devices will most likely not rely on tomorrows technology to get the job done - they're special purpose devices, and as such - stable, open source code - that the developers can tweak and release, giving the device all the functionality it needs and none that it doesn't... is a good thing (tm)
cyn, free software and *nix operating systems enthusiast.
There is a big difference between latest-and-greatest and that which is suitable for the needs of a particular product. After all, what's the point of the latest drivers for the latest whiz-bang device when the device doesn't, and possibly never will, have said device?
If the embedded software used is stable enough and sufficiently featureful, there's really no need to install kernel 2.9.10001030-pre5004-AC234 or whatever.
"That's Tron. He fights for the Users."
Actually... the docs that come with the camera lay out several methods of routing the camera's images through a more powerful web server, so the only thing pulling images from the camera is the web server, which then dishes them out to whomever. I'm not using mine this way (it's not accessible to the outside world), but it should work.
I think, however, the unforseen benefit to unix on a PDA, is just the true insurgence of the business world to unix. If these are cheaper, faster, and freely licensed I think you'll see Admins flocking to these because they can SUPPORT it. Who wants to learn how to fix a messed up palm, or send it to some one?
----
I used to laugh at the iPAQ, echoing the usual "ha! If it's that small, I want a long-battery life palm. Who needs more than that in a palm-top?!" argument, which argument I am not belittling;) [Got a visor last year which keeps on rocking despite drops and battery malnourishment.]
....
However, now I've seen the iPAQ up close a few times (and the itsy, which has a much cooler shape and color!), and I'm 98 percent convinced that iPAQ (or similar) is an actually great idea.
The cons:
- battery life. No getting around it, only choosing better paths.
- Whuddeye miss?
The pros:
- Bright, legible screen. Good enough that with an external keyboard, I'd be happy to use it for a writing station, and OK for it as a web-browsing thing as well, (perhaps with an even slimmer Galeon or a mini-Konqueror?)
- plays mini-movies, MP3s, etc. Which might have been silly a few years ago, but when it's closer to trivial, it's hard to ignore.
- 802.11 via plug-in module - email, in hand, on couch
- Runs X (pro or con, your choice)
In short, I think at $500 (plus accessories) this is coming in very well compared to other tiny computing choices, and even better then those tiny folding keyboards for it are widely avaible.
timothy
jrnl: http://tinyurl.com/c2l8yr / foes: http://tinyurl.com/ckjno5