alt.net.wireless.NOT.FILLED.WITH ANAL RETENTIVE.PRICKS

IME that's exactly the problem with Duh-Link products. They _say_ it does something, but it either doesn't, or it's such a broken implementation that it might as well not be there. Don't get me started. 8*)

Do you have a specific example?

OK, get me started:

DWL-G800AP claims it works as an AP or repeater, and that it can repeat the signal from another DWL-G800AP. Trying to make a laptop roam from the repeater to the AP will hang the AP, stopping all traffic till both boxes are power cycled.

DCS-G900(?) camera has a dyndns.org client in it. It works just fine, but doesn't report it's connected router's IP, just it's own IP. Your "wan" address of your camera will be 192.168.0.42 (OSLT), which will make it hard to get to from the InterWeb.

DCS-G900 camera has a giant security hole. If you know the URL to the password-protected viewing page you can go there directly without the need for a password.

The DP-311U(?) wireless print server has a feature where it'll go check an Email account and print anything it finds. Kind of an Internet fax machine! Wonderful feature, if only it worked.

Tech support is (IME) unresponsive, firmware updates are late and buggy, and they won't accept product returns for broken features. Wonderful feature sets, poor execution. I'll never buy another Duh-Link product, though PT Barnum tells me they'll do really well.

snipped-for-privacy@sushi.com hath wroth:

Anything you can buy today has missing features. In the past, there were routers that did not support WPA. Some are still sold today because there was no FCC order demanding that manufacturers supply adequate security. After most vendors added WPA to their drivers and firmware, WPA2-TKIP was the next missing feature. When the Wi-Fi conglomeration added 3 more methods of authentication, once again many routers lacked those features. Even MS WZC doesn't do all of them.

In the area of convenience features, we have the lack of DDNS support, pre-assigned IP addresses via DHCP, SNMP, logging, diagnostics, statistics, remote admin, stateful packet inspection, rules based packet filters, QoS bandwidth management, wireless client isolation, and such. Many wireless routers have some of these. Very few have all of them.

Incidentally, the local landfill supplied me with plenty of computahs and a few wireless routers. Unfortunately, they stopped accepting computers for recycling and are sending them to metal recyclers.

The web site lists the latest firmware for the SMC7004WBR as dated

1/1/2004. | this an unrelease firmware upgrade? That sometimes happens if the OEM manufacturer releases new firmware for some other vendor, and offers it to other vendors.

I have 4 assorted Belkin products that did not fail, but instead did not function very reliably. I'll spare you how Belkin succesfully wasted my time. Basically, if it doesn't totally blow up, they won't do anything. Two of the items are simple 5 port switches that simply started losing packets. One was an older 802.11b only wireless router that apparently came from the same OEM and apparently uses the same ethernet chip. There was also the not very pleasant experience of finding that the "updated" driver for one PCMCIA wireless card did not work, and only the driver on the original cdrom functioned. There were several reports of others with the same problem in this newsgroup. Calling Belkin support and informing them of the problem resulted in no action. The broken driver is still on the web pile. (Details and model numbers on request if anyone needs them).

We have a different method of operation. The first time I do literally anything, it takes me hours and hours even if it allegedly simple. However, I learn quickly, and the next time, it's simpler. The first time I installed DD-WRT on a Linksys box, it took me about an hour. Learning how to configure it took several evenings. I went through a series of updates, which is now a 10 minute exercise. Note that I usually sell more than one of something, making the learning exercise a requirement.

If ease of setup were a primary requirement, just about anything in networking would be considered a failure. Attempts to automate the processes with UPnP and IRDP (router discovery protocol) have largely failed due to wide range of methods used to connect to the internet. The number of WAN interface methods (PPPoE, PPPoA, satellite, cellular, etc) have increased faster than the manufacturers wizards can adapt. In other words, setup automation has largely been a failure. One could also argue that Cisco, Sonicwall, and 3com make superior wireless hardware, but require substantially more expertise and time to setup.

Linksys and DD-WRT as a different story. People that install alternative firmware do so because they want features and control not available in the stock firmware. In all cases, they are willing to accept the learning curve and setup time in order to obtain these features. Although the average user could certainly install DD-WRT, they will also complain about the added complexity. Fine. Complexity is the price one pays for added features and control.

The emergence and popularity of alternative firmware also underscores a big problem with Linksys. I don't believe their marketting has a clue how their products are being used. The WRT54G v5 fiasco is a great example of how they can kill their own golden goose. Yet, the WRT54G and GS series are their best sellers. They're so good that Linksys recycled the model number on completely unrelated products in the hope consumers will assume it's the same as the popular WRT54G. Maybe they'll realize that its popularity is based on alternative firmware and not anything wonderful about the hardware.

Like I said, none of these bottom of the line vendors are anywhere near perfect.

[POSTED TO alt.internet.wireless - REPLY ON USENET PLEASE]

While I generally agree with much of what you wrote here, my own feeling on this last item is quite different: I care a great deal about not wasting my valuable time on setup problems caused by poor documentation, crappy design, poor quality, and/or poor support. My experience is that such issues strongly indicate that a product isn't "stable, reliable, fast, and bug-free." My time is worth much more than the cost of a better product, so I always use and recommend good stuff -- I usually only get "stuck" when a client or friend buys mostly on price. ("Aren't they all the same?")

John Navas hath wroth:

I assigned a priority to setup time and complexity. I said that I can tolerate considerable complexity and confusion in setup time, if the product is also stable, reliable, and bug-free. Especially, if I only have to do the setup once. As for documentation, if I have to actually read the user docs, then there's something wrong with the product setup and user interface.

While there may be a connection between product quality and ease of setup, I've seen it various ways. Great products with crappy setup, great products with superior setup, and of course crappy products with a crappy setup. However, I have never seen a crappy product with a superior setup. In general, I find that the easy of setup and documentation tends to be several months or revisions behind the actual hardware and firmware.

[POSTED TO alt.internet.wireless - REPLY ON USENET PLEASE]

Indeed -- networking is still Way Too Hard.

My own opinion is that there has been substantial improvement:

• Interoperability has improved

• Products have become more mature and stable

• Products have become more tolerant and less rigid (e.g., smart ports)

• Setup wizards, while far from perfect, are much easier to use

Depends on the target market. Some of those products (e.g., SonicWALL SOHO routers) install easily with little or no expertise.

I don't think that's necessarily the case. I think it's more a matter of maturity, and also of careful engineering. Early in a product life cycle we put up with a lot of crap to get something working at all, but late in a life cycle we expect it to be darn near idiot proof. The problem in 802.11 is a high rate that has kept us early in the life cycle.

I think that's an exaggeration, that much of the popularity was unrelated to alternative firmware.

Yet.

John Navas hath wroth:

I can offer examples where there is a substantial lack of improvement.

- The WEP ASCII to Hex conversion mess is still epidemic.

- I still have compatibility issues with WPA, even with allegedly Wi-Fi certified products.

- The numerous forms of 108Mbit/sec and MIMO are not compatible.

No. They have been replaced by more mature and alledgedly stable products. The products themselves have remained static.

I still see products with an MDI and MDI-X ports for ethernet. Users are constantly plugging a cable into both and wondering why nothing works.

Maybe: Bypassing the wizard in the Netgear WGR-614 requires memorizing an obscure URL. The wizard fails to recognize an internet connection more often than it succeeds.

In general, there is progress. But the progress is invariably with new products, not fixes or "improvements" to existing products.

Gaaak. Art thou serious? The Sonicwall TZ170 SP wireless:

one of the messiest and complicated installs I've ever had to deal with. I still don't have everything working. Punching holes in the security zones for printers was difficult. The policy based NAT was tricky. I suppose setting it up as a basic wireless router would be simple enough. The docs and online help are good. But anything beyond a basic wireless network was a mess. Like I said: If you want lots of features, expect complexity.

Agreed, but not in the case of the WRT54G. Linksys had 4(?) generations of WRT54G that used essentially the same Linux based firmware. Then, they substituted a buggy, inferior (half the ram), and VX Works based piece of junk (v5) that still is having problems. That's a giant step backwards and not anything near my idea of "careful" engineering.

No. The problem with 802.11 routers is that features get added faster than bugs get fixed. Eventually, you end up with a feature infested and bloated router that's full of bugs. DD-WRT is a better example. Every time a new feature gets added, a few *NEW* bugs get introduced. You don't simultaneously get continuous feature development and a stable product.

Oh? Perhaps the return rate on the V5 mutation is an indication of its popularity? I don't have exact numbers but I hear it's rather high.

Never. Features get added faster than bugs get fixed. When the next product arrives from a different OEM, it's back down the learning curve followed by the long uphill climb again. As soon as the bugs are reasonably under control, the next generation product, from a different OEM, are released. Start over.

When the major driving point is lowest price, everything, including quality, suffers. In my never humble opinion, todays wireless routers and access points are no better "quality" than the crude access points I was using in 1997 or so. The difference is that the new ones have far more features, many of which still have bugs.

[POSTED TO alt.internet.wireless - REPLY ON USENET PLEASE]

Don't use WEP.

Rare in my experience. Much lower than WEP problems at a comparable maturity point.

They are compatible at the 802.11g level.

While there are still problems, I nonetheless think interoperability has greatly improved.

Many older products have been improved with firmware updates.

Sure, but things are nonetheless better in the great majority of cases.

Again, I've seen considerable improvement in upgraded firmware.

I'm quite serious. I've installed many SonicWALL products, and they have been easy to setup and configure, by and large, especially as compared to low-end products. You only need to get complicated if you want to do something the low-end products can't do at all. In addition, there is a wealth of documentation to walk you through the more complex issues.

Even good companies can make bad choices. Regardless, I suspect the basic issue there was cost pressure, brought on by a market driven mostly by price for a given feature set; i.e., we probably only have ourselves to blame.

Actually you can and do get less bugs along with more features when development is done properly. "Quality is free."

How high? For what reasons? That's too vague to be terribly persuasive. ;)

Eventually.

When the technology (finally) matures, products generally mature as well, becoming a dependable commodity. V.34 modems are now a case in point.

Again, "quality is free." It really is. See my prior rants on quality. ;)

My own not so humble opinion is that they're lots better today, but do still have a pretty long way to go. I think perhaps you might be frustrated by the fact that so many of the current issues are (a) longstanding, (b) inexcusable, and (c) relatively easy to fix. ;)

Interesting. I've heard the Dlink cameras are to be avoided, but I haven't heard about the repeater problem you mentioned. And tech support did what when you informed them of this?

William P.N. Smith wrote:

snipped-for-privacy@sushi.com top-posted:

IIRC they Emailed me a "here's a couple of hours of stuff to keep you busy, hope you won't bug us again" document detailing all the steps I had already taken (factory reset, reload existing firmware, connect one as AP only, confirm operation, configure second one as repeater, confirm crash, rinse lather repeat). Well, at least they've eliminated me as a drain on their tech support resources.

John Navas hath wroth:

Right. Just junk all the ancient legacy 802.11b only hardware. I have quite a collection of Orinoco Classic Silver cards. Sorry, but WEP isn't just going to disappear overnight. The problem is that the ASCII to Hex conversion mess has been around much longer than WPA. It should have been fixed, standardized, and identified long ago. If you dig through the various support web piles, you'll find nothing on the topic. That's because it almost always happens when a user buys a mixture of equipment from different vendors. The manufacturers don't want to support a different vendors hardware, or deal with hetrogenous systems, so they ignore problem.

Well, I only see the broken hardware. I'm sure that there are many people out there that have a satisfactory out of the box experience, but not when I get involved. Many such issues are fixed by firmware or driver updates. The customers never do these unless assisted or armtwisted by a knowledgeable person. Sometime, I'm in a hurry and just install everything using the CD in box. Then, I fight stupid problems until I remember to update everything.

I'm sure the purchaser of a 108Mbit/sec or MIMO compatible wireless card will the thrilled to know that the added cost of the card is generally wasted because it only works with a compatible access point. Of course, none of this stuff will be compatible with the inevitable

802.11n specification.

As for improved compatibility, methinks you're 99% correct. However, I keep running into card and access point combinations that result in odd compatibility issues. For example, a DWL-G630 card and my BEFW11s4 v4 will contantly connect and disconnet at about a 15 second rate. Both devices work just fine with a variety of other wireless devices, but not this combination.

Many older products have remained frozen with permanent frimware bugs. In general, the way the vendors have fixed such problems is to replace the buggy older product with a completely different model from a different vendor. When the product design cycle is less than the product sales lifetime, there's always a temptation to ignore problems and just offer the customer the new and improved model instead.

I think that's roughly what I said. The simple stuff is simple across the board with most vendors. The added features that justify buying a Sonicwall are not quite so simple.

I know a bit more about what happened with the WRT54G v5. Basically, you're correct. It was a cost decision which overrode any technical and customer use considerations. As I said, Linksys didn't know what their customers were doing with the products. However, I don't think Linksys was trying to be the absolute cheapest. They had a winning product, with the highest sales in the industry, and they (almost) killed it. It's called "pissing in the soup".

That's a great slogan which seems to fail all too often. The problem is not the lack of quality but rather the vendors and customers perception of quality. If they're different, there's going to be problem.

Let's try a different approach. One can maximize their successes or minimize their failures. The approaches to everything from marketing to product development differ radically between these two attitudes. Minimizing one's failures are the "zero defect" programs, which generally fail to produce anything useful. Maximizing one's successes sometime turns into ignoring all problems and losing products and just keep selling the winners. I don't consider either extereme to be useful and a compromise is often necessary. Quality might be free during the development cycle, but once the product has hit the market, damage control is NOT free.

Linksys demonstrated this by raising the price on the replacement V4 and WRT54L models, saturating the retail chain with the v5 models, and following up the mistake with a WRT54GS v6 model, which appears to be more of the same. The first step to solving a problem is to recognize the mistake. I don't think Linksys has done this (yet).

Sorry. That was intentionally vague. All I have are some word of mouth numbers, which I can't leak, and which I'm not even sure are accurate or current.

How many new quality V.34 modem models have appeared in the last 10 years? When the demand slacked off on V.34 modems, the manufacturers had time to clean up their mess. Since there were not going to be any subsequent models, they had a good incentive to fix their problems.

That's not the case with wireless, where the product lifetime is so short, that by the time the problems are fixed, the product is dead. The long delay at Zyxel in fixing known 660HW WPA bugs is one example. The problem has been known for about a year. There's beta firmware floating around that apparently hasn't fixed everything. By the time the firmware is ready for general consumption, the product will be obsolete.

You obviously have more experience in software quality than me. My personal experience is with a marine radio manufactory that considered the measure of quality to be the return and repair rate. If the dealers were returning product, it was presumed that there was something wrong with it. It's my understanding that this is still the case.

Such a measure of quality resulted in some interesting comedies. We were under pressure to deliver product so everything that was deemed non-essential was delayed. That included the service documentation. About 9 months after initial shipments, we were finally able to deliver a proper service manual with test points, schematics, and tuning proceedures. Immediately, the repair and return rate shot up. What had happened was that the dealers were ignoring problems until they could get repair docs. Then, they dived in, and being an unfamiliar product, made a mess. So, they returned it to the factory for exchange or repair. There was a small faction that wanted to eliminate the repair manual as it was directly responsible for the deterioration in "quality".

Anyway, quality may be free, but damage control is not free.

What price do we put on perfection? I had the same problem with SCO. The long development of their operating systems resulted in a substantial number of minor and trivial bugs that never seemed to make the fix list. They were more irritations than bugs and of course, there were always more important bugs to fix and features to add. Eventually, these minor bugs began to accumulate and eventually gave the product a "shoddy" appearance. It was difficult to do anything without running into at least one minor bug.

So, I suggested that instead of working on the major bugs, the company should spend a fixed length of time and concentrate on fixing as many minor bugs as possible. The logic was that 1000 such minor bugs could be fixed in the same time frame as one major bug or feature addition. I convinced almost everyone except those making the decisions, so the idea was dropped.

Now I get your point. My problem with "missing features" is when the manufacturer says it will do X,Y, and Z, then you buy it and find out it only does X andY, but they will supply a patch to do Z. Your idea of a missing feature, assuming I am interpretting you correctly, it that the box that does X and Y should do Z if the marketing team didn't have their heads up their ass.

Was this an unrelease firmware upgrade? That sometimes happens if the

SMC released a firmware upgrade for the box, then believe it or not, just recalled it. No new firmware upgrade, but just go back to the old one. Now I can see fixing the fixes, but going back to the old rev makes me think there was something worth fixing, but they just couldn't get it right. Needless to say, I'm not buying SMC gear again.

I've been plagued with the router just locking up once every few weeks. Before throwing it in the trash, I found that later rev and the problem seems to be gone. I'd like to replace it with a gigabit router, but they don't seem to exist. You can find gigabit switches, but nt set up for a wan.

I only install gear for myself, while I gather you have a business of sorts. Thus the first install I do is the last one.

[POSTED TO alt.internet.wireless - REPLY ON USENET PLEASE]

Yep. Sometimes painful (albeit not terribly expensive), but yep. Lots of other attendant benefits (e.g., stability, features, compatibility, support, performance, ease of use).

I've filed all mine away. :)

True, but I'm doing my best to speed it along. :)

Like lots of other things.

I think it's now more a matter of the Wi-Fi Forum not having much if any interest in a minor band aid to an embarrassment they'd rather forget. It's arguably counterproductive and a disservice to fix it.

I'd be willing to bet that the majority of people (a) don't even notice or (b) accept same-vendor as a reasonable evil.

I'm reminded of a friend that asked me by-the-way if I could fix her poor broadband and Wi-Fi. I found she was actually unknowingly using a neighbor's broadband (thanks to multiple screwups by paid "Geek Squad" installers), and the poor performance was due to weak signal. (In well over a year she'd never even used the broadband she'd been paying for.) It took but a few minutes to reconfigure her gear properly (including WPA) so her broadband works wonderfully well. She thinks I'm 'a genius." ( If only. ;)

Both are true, but that's still a measure of progress. Those who buy on price alone tend to get what they deserve.

Also: "Messing with a good thing." "If it ain't broke, don't fix it."

With all due respect: (a) The customer (market) perception is the only one that really matters. (b) Proper quality procedures are well-defined and documented. (c) There are many, many proofs that in fact "quality is free." I'll go even farther and say that a proper quality process actually results in lower cost and faster delivery. That so many companies do it wrong doesn't change that.

I strongly disagree. The common misconception is that you have to test quality into a product, a process that never really reaches closure, in part because new bugs are introduced by the fixing of old bugs. You can't test quality into a product -- that simply doesn't work. What's needed is a *total quality process* from the very beginning through to the very end. Documentation can and should be written *before* the code is written. That way written code tends to *work right the first time*. (Lots more.)

That's essentially throw-it-against-the-wall-and-see-what-sticks, which is inevitably a recipe for failure in a competitive market.

Quality *is* free for that very reason: If it costs $1 to fix something in the design process, it costs $10 to fix it in the implementation process, $100 to fix it in the testing process, and $1000 in the field -- the cost of support is another reason that quality

*is* free.

"More of the same" is another all too common mistake when something isn't working, especially in this country. 'Incarcerating the highest percentage of the population in the industrialized world hasn't stopped violent crime, so let's raise the percentage ("get tough on crime")!'

I think that's what I'm saying. ;)

Because the market is still immature. As it matures, things will get better.

Yet another misconception, compounded by calling it "the cost of quality."

I'm not surprised. Yet the great majority of most returns when analyzed are "no trouble found." Meaning the manufacturer doesn't really understand what the market expects and needs.

True, yet another reason that quality *is* free.

To be clear, I'm not talking "perfection," just meeting market needs with a quality product.

While that may sound appealing, in practice it's rarely effective, if for no other reason than the introduction of new bugs as part of the fixing of old bugs, which tends to be more frequent the older they are, resulting in an unacceptably long and expensive process.

It's much better to put in place a *proper quality process* before embarking on bug fixing. That might entail a fair amount of code re-writing, but that may well have lower total cost than (a) leaving bugs unfixed or (b) patching them poorly.

[POSTED TO alt.internet.wireless - REPLY ON USENET PLEASE]

What if they never acknowledge the missing feature, or claim it does Z for small values of z, or claim it works in their lab, so you must be doing something wrong, and they don't accept returns?

Why would they exist? What are they good for? The fastest SOHO broadband I'm aware of is 30 megabits, why do you need more than

10/100 on the WAN port? [If you mean you want gigabit on the LAN side, then feed the output to an appropriately-sized gigabit switch, they are pretty cheap now.]

I've got 3 10/100 switches, a PoE switch, three routers, three gigabit switches, two APs, (and a partridge in a pear tree) in my infrastructure. Having a few extra gigabit ports on the LAN side of my internet router wouldn't really be very useful, so maybe I'm just missing the utility of it all.

Yeah, I see the cheap (Airlink) gigabit switches at Frys, and also those from major brands. It just seems like I shouldn't have to add a switch (with associated headaches), when it must cost didly squat to make new routers with high speed lan and also DSL WAN interface.