Monday, August 31, 2009

The sliver bullet

As I said earlier, it is nearly impossible to make products without some problems. The more products you connect together, the farther away the time sensitive data has to travel, the more data being transported and the higher the quality has to be, the more problems you are going to run into. In this case, we are talking about high definition TV in a home network which means a lot of data which has to be transported flawlessly or the average consumer can see the glitch. Also, add to this the fact that the DLNA solution is a consumer retail solution and you have to do everything well at a very low price.

It is too early to declare DLNA a failure, but I would not be surprised to see them fizzle. In my mind, the problem is with the channel. You can not start out with something this complicate by going retail, selling to the guy and gal who could barely program their VCR. You need to start this sort of thing by selling to the professional home network installer.

The professional installer is trained on systems, skilled at workarounds, technical and they will find the products that work together and only use those products. They do not randomly go to Wal-Mart and pick up a Colby this to work with an Apex that. Retail is an interoperability nightmare for anything complicated. Installers find what works and stick with it as much as they can.

Once they install a system, their customer is asking for a service call if they tinker with it. This is totally different than a system the customer buys at the store and installs by themselves. The silver bullet for HD home networking is to go through a professional until you know what you are doing. Then enter retail with a few products and grow the market from there. DLNA tried to rush into retail with dozens of products from dozens of companies. You just can not test for everything that will happen once the product leaves the store.

Sunday, August 30, 2009

The Competiton Slips

Ethernet was never intended to carry time sensitive data, such as video. With video, every frame needs to make it to the TV screen on time or the results are unacceptable. DLNA had to do a total rework of Ethernet. The result is a new technology based on Ethernet but one that will not work with traditional Ethernet products.

1394, on the other hand, was developed to carry time sensitive data and to prevent collisions.

All this left me wondering why they would do a massive rework of an ancient technology (Ethernet), then rebrand it as “DLNA” so that only “DLNA” products would be expected to work together (and no Ethernet products). It would have been faster to have added a few features to 1394 and called it “HANA” (for example) and then built expectations that HANA products would work with HANA products. The nonsense of their logic was stunning.

Two years ago, the DLNA group hired a full time person to oversee their compliance program. However, even with this level of effort to assure interoperability, their product introduction failed. Earlier this year, retail giant Best Buy rejected DLNA claims to interoperability after problems surfaced in their own private tests. Several DLNA participants admitted in the press that the group had failed to deliver as promised.

Saturday, August 29, 2009

Samsung...exit stage left

Home networking over 1394 seemed like it should be a slam dunk. However, several notable companies and organizations had already tried and failed. Currently, there is one last attempt being developed, but the outlook is bleak. If 1394 home networking has so much promise, why can it not gain any traction?

The interoperability issues are a big problem. Samsung decided to abandon their 1394 home networking effort because old 1394 devices, when connected to a 1394 network, will cause bus reset storms, locking up the system. They were worrying about products that were made 10 years ago, most of which are no longer being used. I encouraged them to establish a strong compliance testing program and only guarantee their products to work with products that passed the certification tests. But their response was that they were afraid that customers would see the 1394 port and try to connect the old devices anyway.

Large consumer A/V companies can be frustratingly paranoid to deal with, being painfully slow to adopt new technologies. You have to make them worry that they are being left behind by the competition before most of them will inch forward.

Surprisingly, after backing away from 1394 for networks, Samsung threw their effort into an Ethernet based home networking initiative called “DLNA”. They went straight from the frying pan into the fire.
Ethernet was developed for mainframe computers in the early 70s (patented in 1975). It was developed to connect every device that wanted to gain access to the network. When everything is connected with equal access to the network, bad things happen. Fully loaded networks can get very busy and very slow. If you’ve ever worked in an office with a shared printer accessible through a network, you’ve heard the printer pause for a second or two as it waits for the next packet of data to come across the busy network. Ethernet will not prevent data from colliding on the network. It manages the collisions. Sometimes this requires data to be sent again. Not a big deal when printing a document. A few pauses along the way will not seriously slow down the print job. But imagine watching a bowl game with the TV screen pausing two seconds each time the network gets busy. The store returns on such a product would be record breaking.

Despite a rapidly growing list of member companies joining their trade association, storm clouds were forming across the DLNA landscape...

Friday, August 28, 2009

Another Year, Another Revolution... that fizzled

Retail home networking is price sensitive, but professionally installed home networking is not nearly so price sensitive. I see HDMI 4X4 matrixes selling on-line for $1,000. They have very little inside and do very little. With FireWire’s capabilities, a home network could transport five different streams of high-def video (with Dolby audio) to five HDTVs across one single Coax or CAT5 cable, moving video in both directions and be controlled by a graphical user interface sending control signals across the same wire.

I should repeat that again slowly because that is a lot to take in. Imagine five different devices, such as an HD Cable Box, a Blu-Ray player, a Digital Video Recorder (a TiVO like device), a central media controller with on-line access to Netflix, and a computer connected to the internet with family photos and home movies on it. That is five different video sources at a minimum. If the HD Cable Box had two tuners in it, then it would be able to provide two different HD channels at the same time…making it essentially two video sources. If it also had a PVR inside, that could provide an additional option for video content. Now imagine sitting in front of an HDTV in your living room, hitting a button on a simple remote control (power, menu, up arrow, down arrow, select). Select “menu” and a menu comes up showing all the video sources connected to the network. Select one, HD cable box (for example) and you can pull up a new menu to show you a full program guide or you can select the PVR and see everything that has been recorded on the PVR. Hit “menu” again and you can see what DVD is in the Blu-Ray player or what videos have been stored on the Digital Video Recorder and select anything you want to watch. All this with a five button remote. You have access to all the content in your home available to any TV in the living room, kitchen, bedroom, media room, pool, garage, or any place you care to run a cable. Not just one TV, but five TVs can play any video stream they want at the same time.

At a $1,200 - $2,000 price point ($1,000 for the full function media server with a $200 access box at each TV), a feature rich network like this would sell well in retail and would revolutionize the professional installer industry.

Why have we not seen this, because a device like this was developed and withered due to lack of interest? It fizzled for four reasons. Reason number four was the current economic downturn that dried up investment dollars and made companies reluctant to branch out in bold new directions. The other three reasons will be covered in my next installment.

Thursday, August 27, 2009

Our Strength was our Weakness

I was discussing the incredible business opportunity I see in home networking over 1394 with a German 1394 enthusiast when his comments brought me up short. This is a person who has built a company doing large professional networks with 1394. Originally he was trying to use Ethernet but found the limitations of Ethernet made it unsuitable for large networks carrying real-time data such as professional recording studios or high definition video distribution. He was increasing buffers and creating sub-nets and all sorts of other things to get Ethernet to work. The so called “inexpensive Ethernet solution” so commonly referenced in networking discussion was costing a fortune to debug. He switched to 1394 and found that 90% of his problems went away. The chipsets were more expensive but his ability to do what he needed to do without taking heroic measures made 1394 a cheaper solution.

I expected him to embrace this new and growing market; home networking. To my astonishment, he indicated that 1394 was not good for cost sensitive solutions.

The fact that 1394 is better suited to complex applications is also its limitation. 1394 is much more capable than USB, Ethernet, HDMI, and a whole host of other choices. In reality, it does not cost that much more to make. I have heard, for example, that a USB cable costs around 75 cents to make and a FireWire cable around a dollar and HDMI just over a dollar. That should cause you some irritation the next time you pick up a cable at Best Buy, but there are a lot of things that go into setting the retail price than simply the cost of making the product.

The issue is not so much the cost of 1394 as much as it is the cost of developing the solution that needs 1394. You do not need 1394 in a mouse or keyboard. Even USB is overkill for those devices and since it is a few cents cheaper, USB wins that socket. When it comes to printers and scanners, the same holds true. USB is adequate for so many things, and being a few cents cheaper makes it the preferred solution.

The system being developed needs to be very complex before it will require the capabilities of 1394. Those systems are expensive to make. It is not the price of 1394 that makes it expensive. It is the capabilities of 1394 that make it only suitable for expensive solutions. In audio equipment, for example, the cheap consumer products use USB for a point-to-point connection to a computer. For professional systems with multiple streams of audio going between equipment along with command-and-control being transported back and forth, 1394 wins the socket.

This was a novel idea to me. I’d always promoted FireWire as a better all around solution, able to do everything, never realizing that our strength was also our weakness.

Wednesday, August 26, 2009

Doing it right and still not working...

Problems will emerge when making a product no matter how big you are or how long you have been in the business. Once a company realizes they have a problem with the FireWire port on a device, they are not going to publicize it. They will make sure that all their products work together and then hope for the best. Compliance testing, it was feared, would only make it obvious that Company A was having design issues. To remove the stigma of being one of the few companies testing products, the 1394TA worked with member companies to get everyone to start testing products once the tests were available. However, the feedback from major companies was that they were not getting significant returns due to FireWire bugs. Additionally, to start compliance testing 7 – 8 years after products started shipping could only serve to raise concerns. It was speculated that customers would assume that significant problems had only recently been discovered and suddenly the whole industry was franticly testing products. The whole effort was sunk from the beginning.

But there was an additional issue which caused fully compliant products to fail when connected over FireWire. The 1394TA standards included a few mandatory features and a long list of optional features. If Sony decided to support optional feature X and Panasonic chose to not support that option, then feature X would not work between the two products. Imagine a dozen features that may or may not be supported on all the products in your home and you can see the problem.

Compliance testing would do nothing to remedy this problem.

1394/FireWire was not alone in this. Recently, retail giant Best Buy failed products using a competing technology (DLNA) largely due to the “too many optional features” problem. To find an article on this issue, simply Google “Best Buy DLNA”.

In the computer world, these problems were largely transparent to the user. The operating systems from both Apple and Microsoft mask most problems. The software simply included 100s of work-arounds to prevent the products from failing. Once again, 1394 is not alone in this. USB developers have frequently stated to me that 1,000s of bug fixes are in the operating systems from MS and Apple to keep USB working. Those who are not mouthpieces for the industries making money off of USB (i.e. people not selling USB silicon or cables) decry the poor quality of the USB technology and the millions of man hours required to get it working and keep it working. But it works “good enough” and that is all the consumer knows.

There is a lot to be said about “good enough”. 1394 could have learned a lesson there as I will cover in the next installment.

Home Networking...the plot thickens

In addition to intentional judgment calls that lead to inconsistencies between products, there are myriad tiny design mistakes which are legion with 1394 products. Even among household names, dozens of common design mistakes are all too common.

For example, if your board layout includes differential traces which are different lengths, you can introduce skew. I.E. if the board in your product with the computer chips on it uses wires of different lengths, you can cause some of the audio, video, or data to arrive a little bit late.

Even more shocking (pun intended), many manufactures do not design their product to accommodate the right electrical current carried by the FireWire wire. Hook up a device providing the full FireWire electrical current and the poorly designed product will not be able to withstand the current coming over the FireWire cable.

The power problem would produce a more obvious failure than the board layout problem. Most of the common design mistakes are not so obvious. For example, manufactures sometime do not understand some of the information their product needs to provide when communicating with other products over 1394. They will leave information out or duplicate information in all their products causing significant confusion and poor performance.

These problems could have been significantly reduced if the 1394TA had a strong compliance program. That is great in theory, but difficult in reality. The first 1394 product emerged in 1995 but the first meeting to discuss a compliance program did not take place until 1998. Meaningful work did not start on a compliance standard until 2002. After 7 years of products shipping, the sense of urgency for testing products was very low and the TA was powerless to do anything about it.

Monday, August 24, 2009

Home Networking...fallen and can't get up

After a great start, home networking over 1394 fizzled…repeatedly. After several years of analysis, some answers emerged.

For one thing, companies tend to follow standards imprecisely. A standard needs to be precise enough to assure that products will work together but you can not cover everything. If you even try, you end up with a standard that takes more years to write than the useful lifecycle of the technology. In the world of technology, the better mouse trap is constantly being introduced and if the mouse trap gains market acceptance, all old versions quickly become obsolete.

Therefore, there are subtle differences that can be introduced by well intentioned product designers when they use their best judgment about what makes sense when the standard is not specific. These judgment calls can make a Sony product fail intermittently when used with non-Sony products.

I mention Sony because they got a bad name in the industry for going their own way and deviating from the standard when the standard did not meet their need. Even worse, they were accused of intentionally introducing subtle differences in their products to encourage customers to buy all Sony products instead of buying products from multiple vendors.

Admittedly, Sony did deviate from the standard as they did when they introduced the 4-pin connector however; it is my opinion that they did not introduce subtle flaws to intentionally make their product work poorly with those of their competitors. Such a strategy would be too dangerous for the reputation conscious Sony I know. The customer could just as easily blame the Sony product for the failure as he could a non-Sony product. Additionally, I am aware of work Sony and Panasonic did with their competitors to make sure that Japanese products worked well together.

Sony was the leader in 1394 products by a large margin. As with any leader, when things go wrong, they were an easy target for cheap shots, paranoia, and conspiracy theories. I think this was simply a case of Sony making their best judgment call and simply not meshing on every point with the competition.

However, there was a more obvious problem which led to incompatibilities and the problem came straight from the standards written by the 1394TA.

Saturday, August 22, 2009

Home Networking

About 5 years ago, the 1394TA marketing working group chairman and I decided that we needed to position 1394 as a unique solution, able to do something that nothing else can do. This is consistent with item two in my “Apple formula for success” list. We selected Home Networking. USB can’t do it. HDMI can’t do it. At the time, DLNA was not around, so Ethernet could not do it. Home computers had 1394 as did digital STBs, Digital VHS tape decks, and a growing number of HDTVs.

The idea of using 1394 for home networking had been around for years. Gary Hoffman (first chairman of the 1394TA) presented 1394 as a home network at the first 1394 Developers Conference on July 28, 1997. VESA Home Networks (now the DisplayPort people) announced a “proof of concept” using 1394 on April 1, 1998. CEA showed the Virtual Home Network (VHN) based on 1394 at CES on January 8, 2001. When USTec announced tecStream with 1394 in April of 2005, things looked good, but they pulled the product about a year later due to “interoperability problems working with 1394 products in the field.” No VESA network, no VHN network, no tecStream network…

What happened?

And what is happening now?

More to come....