One does not need to hack a wireless card's firmware to screw up the 2.4GHz ISM band. There are many ways to spew RF at the correct frequencies. (every home in the nation has the one device needed to do it:-))
(In fact, I have 2 ancient Breezenet PCMCIA "802.11" cards right here that will quite happily stomp all over more modern 802.11b/g gear -- while still being fully functional.)
Negative. They are specifically mandated as per the coupon program to be ATSC receivers only. While I am unaware of any digital tuning hardware that doesn't support both ATSC and QAM, the rest of the box (hardware and software) has to know what to do with the data.
If this were true, then my HR10 Tivo should be able to tune digital cable channels, but it most definately doesn't. (It doesn't tune OTA very well either -- early generation tuner with ZERO multipath handling.:-))
The exact ruling stipulates that cable operators must make local stations available to the "basic teir" (i.e. lowest level.) For any cable company still offering analog cable, that means converting the now digital broadcasts to analog for transmission. If their network is 100% digital, they don't have to convert anything... you'll need a STB or a "Digital Cable Ready" TV. In just about every cable market, it's a trival cost to add converters for the roughly dozen local stations, but the FCC had to put it on paper to force the greedy 'tards to actually do it. ("cheap" isn't free, so they generally aren't motivated to do it.) I, for one, would like to see analog cable gone entirely, but there are just too many subscribers.
It's MPEG-2. But yes, they can, and do, shoespoon a dozen SD channels on one carrier ("QAM"). 256QAM provides just shy of 40Mbps per channel -- exactly 2 full ATSC streams, so 3-4 HD channels will usually fit. (I've seen TWC push 6 per channel, so beware of HD-Lite.)
The reason for the continued existance of analog cable is the number of people still paying for it. It has nothing to do with set-top hardware as no one has needed a "cable decoder" for a decade -- analog scrambling went away the day digital cable sprang into existance; the difference between basic cable and standard cable is a physical filter on the line.
Hah. No. Cisco and Juniper are both out of touch with reality.
The new Cisco 2851 hanging on the wall with a DS3 run into it is actually slower than the PIX 520 (decades old now) in the rack next to it. Granted, by Cisco's own documentation, it's not rated for a full DS3, but even at a fraction of the speed, it cannot handle the NAT, IPSec, and routing the pix has been doing for years. 2851 is at 70%+ util while the pix peaks at 1%. And the 2851 has hardware crypto support, so don't think the less-than-T1-rate VPN traffic is the cause. If you want an expensive device to copy packets from one interface to another (until the end of time), Cisco's got you covered.
You don't need 6 nor do you need fiber. 24 port gig switches aren't that expensive these days. (Netgear, HP, Linksys/Cisco... it's all the same broadcom crap under the hood.)
You still didn't answer the question... what exact did you replace? 10m hubs running on 10base-2?
And what was their cheap gear you replaced that was so much slower than the slow cisco 2924's? I'll go out on a limb and guess you replaced HUBs with SWITCHES; that's an improvement no matter how cheap the switch. For $450 you could go gigE these days. (with NEW gear even.)
I would upgrade it to Linux and Samba but the adaptec raid card has no stable drivers for Linux.
dpt_i2o? I've used one for a long time. Yes, I had to fix the driver DPT published (a SCO driver hammered into Linux and FreeBSD) -- this was before Adaptec took over and put A person incharge of the driver.
Everything he listed is very long past EOL. Cisco WON'T support you, no matter how much money you may be willing to throw at them. Even better... getting used equipment that's still under someone else's support contract. (unless you have an inside man at Cisco, that one rarely gets fixed.)
Cisco *used* to make great hardware. Today, they are quite a bit behind the times and even more expensive -- who uses 100M ethernet gear these days. Factor in their EOL policies that force you to move to new equipment every few years, and it's even worse. I know a lot of places that still use 2500's and are perfectly happy with them.
Exactly. Taking the platters out of the drive doesn't alter the data on them at all. So, he's done less to protect that (worthless) data by turning it into a tree.
And DoD "erasure" is basically "destroyed with thermite"; any drive that has ever held classified data is physically destroyed, not erased. I will never get over people's blind stupidity over "data security"... zeroing the drive is more than enough to erase your data from all but about 10 people on Earth. Trust me on this one: Your. Data. Isn't. THAT. Valuable. (and to be honest, most of the stuff labeled "classified", isn't worth the effort either.)
xtrek may seem primitive compared to modern games, but it was the shit in it's day. And it was a scourge on campuses all over the globe. At NCSU it was disliked by the admins because people tied up public workstations for hours playing when other people needed to do their classwork. More than once I had to run people out of lab to have seats for all of my students -- they prefered those machines because the ops would have to walk behind them to see they were playing. AND, the admins had a box full of dead ($100) DECStation mice that had been killed by the relentless, non-stop clicking of xtrek players.
Now, with modern consumer electronics, it is equally trivially possible to copy your entire movie library...
Not exactly. The digital outputs are "protected" to prevent just such copying. The analog outputs are not in HD. The component HD outputs take expensive equipment to capture, and the DRM can prohibit analog HD output entirely. (it's coming along, but still not cheap or readily available.)
Funny. I built my own "stabilizer" using junk recovered from various broken TVs, radios, etc. I used a slow opamp, so it killed the color burst as well, but for $0 it worked just fine. I rigged another VCR to be immune to the BS (disabled it's AGC circuits.) And my Go dual deck VCR doesn't care at all about macrovision -- it'll duplicate any tape at about 4x normal speed, macrovision and all.
In the hundreds. If you go back far enough, they were over 1000$ for a while. VTPs (video tape players) were cheaper because they didn't record -- despite being 99% identical to those that could. (the industry didn't want you to be able to record anything. It's the same sad song they've been singing since the vinyl days.)
This isn't a limit; it's a static size. Having the system allocate memory to support what it needs is a far better system than being a lazy coder and creating a ton of fixed size tables. Over the last 20 odd years, I have seen code from far too many (stupid) lazy "programmers" who create fixed size structures for things that do not have a fixed size. It's a combination of lazy programming reinforced by every improving technology. Why should we spend the time writing fast, efficient code when we can "just through more GHz at it." The same is said for memory... why use memory efficiently when the machine has multiple GB of really fast memory.
(ASICs aren't exactly little tiny CPUs with their own OS to manage alloc() and free(), so they have to have bits carved in stone -- unless they are FPGAs that can be reprogrammed on the fly.)
We're talking about software here. The hardware's limit is something very different -- software cannot make the hardware buffer any larger. I've seen "tricks" where the hardware switches the most active traffic (or the first MACs to get in the table) and then process switch the rest. And I've seen cheap (ala netgear, but everyone uses the same broadcom crap) switches that either stops passing new traffic, or starts unicast flooding packets like mad.
(However, it's been a long time since I was around a network that would fill a 4k mac table. And then it was a 10base-2 network.)
One word, on one line, in one file... but you have to know which file. C vs. C++ doesn't matter. It's all a matter of knowing the codebase.
The real WTF is why code a fixed size for something that is dynamic by nature. The MAC table may need 5 entries, and it might need 5000. (I had this *cough*discussion*cough* with a bunch of pure idiots at 3com 15+ years ago... hardcoded, static sized route tables. I have not bought a single piece of 3Com crap since then.)
The code is incomplete and not current. Sadly, this is a near universal problem with commercial use of linux. Netgear is horrible in this respect -- they put broken tarballs up and then take months to correct it. All things concidered, Linksys is pretty good at it -- they rarely post 100% of the source, but across many releases you can find 99% of everything, and they don't take months to put up the source for current firmware releases.
(What 3rd party devs would love to see is current source for the wireless drivers. The last broadcom source that was "leaked" was years ago -- I don't think linksys intended to include *that* directory in the tarball.)
Actually, there are several Cisco products that have Linux under the hood. Most notablly the ASA line of firewalls -- and there's no mention in anything Cisco provides to say it's running Linux, or offer source code, nor do they hand over any source if you ask them for it. In this case, it's just the kernel; the entire thing that makes it a cisco firewall is the single application ("lisa") it runs.
Actually, Linksys did have "business class" hardware (WAP4400, etc.) that was (is) much more expensive. But I still wouldn't call it "enterprise" hardware.
But don't be so quick to jump on Cisco. They bought this problem when they bought Linksys. Linksys has been doing this half-assed source code BS since the beginning of time.
First off, Cisco bought a company to get the "aironet" stuff, just like they bought Linksys and landed in this mess. If they could get ANY IOS image to fit in 2M flash, and run in less than 8M RAM, I'm sure all the little linksys trash would be using it too. The enterprise hardware isn't as price sensitive as the consumer/retail market, so they can afford to go a bit overboard with the hardware.
They went back to VxWorks. Their claim is it's smaller, so they can put smaller (cheaper) flash chips, and less RAM in the systems. But I seriously doubt the cost savings amounts to what they have to pay Wind River in licensing.
Both OpenWRT and DD-WRT produce fully opensource firmwares for a number of systems, including the crippled 54g v5+. All of my linksys toys run dd-wrt (well, except the uber-ancient WAP11, but it's just an atmel microcontroller with a PCMCIA wireless card bolted to it.)
Slashdot Mirror
One does not need to hack a wireless card's firmware to screw up the 2.4GHz ISM band. There are many ways to spew RF at the correct frequencies. (every home in the nation has the one device needed to do it :-))
(In fact, I have 2 ancient Breezenet PCMCIA "802.11" cards right here that will quite happily stomp all over more modern 802.11b/g gear -- while still being fully functional.)
Negative. They are specifically mandated as per the coupon program to be ATSC receivers only. While I am unaware of any digital tuning hardware that doesn't support both ATSC and QAM, the rest of the box (hardware and software) has to know what to do with the data.
If this were true, then my HR10 Tivo should be able to tune digital cable channels, but it most definately doesn't. (It doesn't tune OTA very well either -- early generation tuner with ZERO multipath handling. :-))
The exact ruling stipulates that cable operators must make local stations available to the "basic teir" (i.e. lowest level.) For any cable company still offering analog cable, that means converting the now digital broadcasts to analog for transmission. If their network is 100% digital, they don't have to convert anything... you'll need a STB or a "Digital Cable Ready" TV. In just about every cable market, it's a trival cost to add converters for the roughly dozen local stations, but the FCC had to put it on paper to force the greedy 'tards to actually do it. ("cheap" isn't free, so they generally aren't motivated to do it.) I, for one, would like to see analog cable gone entirely, but there are just too many subscribers.
It's MPEG-2. But yes, they can, and do, shoespoon a dozen SD channels on one carrier ("QAM"). 256QAM provides just shy of 40Mbps per channel -- exactly 2 full ATSC streams, so 3-4 HD channels will usually fit. (I've seen TWC push 6 per channel, so beware of HD-Lite.)
The reason for the continued existance of analog cable is the number of people still paying for it. It has nothing to do with set-top hardware as no one has needed a "cable decoder" for a decade -- analog scrambling went away the day digital cable sprang into existance; the difference between basic cable and standard cable is a physical filter on the line.
Hah. No. Cisco and Juniper are both out of touch with reality.
The new Cisco 2851 hanging on the wall with a DS3 run into it is actually slower than the PIX 520 (decades old now) in the rack next to it. Granted, by Cisco's own documentation, it's not rated for a full DS3, but even at a fraction of the speed, it cannot handle the NAT, IPSec, and routing the pix has been doing for years. 2851 is at 70%+ util while the pix peaks at 1%. And the 2851 has hardware crypto support, so don't think the less-than-T1-rate VPN traffic is the cause. If you want an expensive device to copy packets from one interface to another (until the end of time), Cisco's got you covered.
You don't need 6 nor do you need fiber. 24 port gig switches aren't that expensive these days. (Netgear, HP, Linksys/Cisco... it's all the same broadcom crap under the hood.)
You still didn't answer the question... what exact did you replace? 10m hubs running on 10base-2?
And what was their cheap gear you replaced that was so much slower than the slow cisco 2924's? I'll go out on a limb and guess you replaced HUBs with SWITCHES; that's an improvement no matter how cheap the switch. For $450 you could go gigE these days. (with NEW gear even.)
dpt_i2o? I've used one for a long time. Yes, I had to fix the driver DPT published (a SCO driver hammered into Linux and FreeBSD) -- this was before Adaptec took over and put A person incharge of the driver.
Everything he listed is very long past EOL. Cisco WON'T support you, no matter how much money you may be willing to throw at them. Even better... getting used equipment that's still under someone else's support contract. (unless you have an inside man at Cisco, that one rarely gets fixed.)
Cisco *used* to make great hardware. Today, they are quite a bit behind the times and even more expensive -- who uses 100M ethernet gear these days. Factor in their EOL policies that force you to move to new equipment every few years, and it's even worse. I know a lot of places that still use 2500's and are perfectly happy with them.
Exactly. Taking the platters out of the drive doesn't alter the data on them at all. So, he's done less to protect that (worthless) data by turning it into a tree.
And DoD "erasure" is basically "destroyed with thermite"; any drive that has ever held classified data is physically destroyed, not erased. I will never get over people's blind stupidity over "data security"... zeroing the drive is more than enough to erase your data from all but about 10 people on Earth. Trust me on this one: Your. Data. Isn't. THAT. Valuable. (and to be honest, most of the stuff labeled "classified", isn't worth the effort either.)
I have a machine that holds 800 discs. So I don't have to do anything but push a button to change discs. Yeah, it's a lot bigger than a hard drive.
xtrek may seem primitive compared to modern games, but it was the shit in it's day. And it was a scourge on campuses all over the globe. At NCSU it was disliked by the admins because people tied up public workstations for hours playing when other people needed to do their classwork. More than once I had to run people out of lab to have seats for all of my students -- they prefered those machines because the ops would have to walk behind them to see they were playing. AND, the admins had a box full of dead ($100) DECStation mice that had been killed by the relentless, non-stop clicking of xtrek players.
Have you ever tried doing that? The results are not worth the time and effort.
Not exactly. The digital outputs are "protected" to prevent just such copying. The analog outputs are not in HD. The component HD outputs take expensive equipment to capture, and the DRM can prohibit analog HD output entirely. (it's coming along, but still not cheap or readily available.)
Funny. I built my own "stabilizer" using junk recovered from various broken TVs, radios, etc. I used a slow opamp, so it killed the color burst as well, but for $0 it worked just fine. I rigged another VCR to be immune to the BS (disabled it's AGC circuits.) And my Go dual deck VCR doesn't care at all about macrovision -- it'll duplicate any tape at about 4x normal speed, macrovision and all.
In the hundreds. If you go back far enough, they were over 1000$ for a while. VTPs (video tape players) were cheaper because they didn't record -- despite being 99% identical to those that could. (the industry didn't want you to be able to record anything. It's the same sad song they've been singing since the vinyl days.)
This isn't a limit; it's a static size. Having the system allocate memory to support what it needs is a far better system than being a lazy coder and creating a ton of fixed size tables. Over the last 20 odd years, I have seen code from far too many (stupid) lazy "programmers" who create fixed size structures for things that do not have a fixed size. It's a combination of lazy programming reinforced by every improving technology. Why should we spend the time writing fast, efficient code when we can "just through more GHz at it." The same is said for memory... why use memory efficiently when the machine has multiple GB of really fast memory.
(ASICs aren't exactly little tiny CPUs with their own OS to manage alloc() and free(), so they have to have bits carved in stone -- unless they are FPGAs that can be reprogrammed on the fly.)
We're talking about software here. The hardware's limit is something very different -- software cannot make the hardware buffer any larger. I've seen "tricks" where the hardware switches the most active traffic (or the first MACs to get in the table) and then process switch the rest. And I've seen cheap (ala netgear, but everyone uses the same broadcom crap) switches that either stops passing new traffic, or starts unicast flooding packets like mad.
(However, it's been a long time since I was around a network that would fill a 4k mac table. And then it was a 10base-2 network.)
One word, on one line, in one file... but you have to know which file. C vs. C++ doesn't matter. It's all a matter of knowing the codebase.
The real WTF is why code a fixed size for something that is dynamic by nature. The MAC table may need 5 entries, and it might need 5000. (I had this *cough*discussion*cough* with a bunch of pure idiots at 3com 15+ years ago... hardcoded, static sized route tables. I have not bought a single piece of 3Com crap since then.)
The code is incomplete and not current. Sadly, this is a near universal problem with commercial use of linux. Netgear is horrible in this respect -- they put broken tarballs up and then take months to correct it. All things concidered, Linksys is pretty good at it -- they rarely post 100% of the source, but across many releases you can find 99% of everything, and they don't take months to put up the source for current firmware releases.
(What 3rd party devs would love to see is current source for the wireless drivers. The last broadcom source that was "leaked" was years ago -- I don't think linksys intended to include *that* directory in the tarball.)
Actually, there are several Cisco products that have Linux under the hood. Most notablly the ASA line of firewalls -- and there's no mention in anything Cisco provides to say it's running Linux, or offer source code, nor do they hand over any source if you ask them for it. In this case, it's just the kernel; the entire thing that makes it a cisco firewall is the single application ("lisa") it runs.
Actually, Linksys did have "business class" hardware (WAP4400, etc.) that was (is) much more expensive. But I still wouldn't call it "enterprise" hardware.
"We're a multi-billion dollar company. F*** off!"
But don't be so quick to jump on Cisco. They bought this problem when they bought Linksys. Linksys has been doing this half-assed source code BS since the beginning of time.
First off, Cisco bought a company to get the "aironet" stuff, just like they bought Linksys and landed in this mess. If they could get ANY IOS image to fit in 2M flash, and run in less than 8M RAM, I'm sure all the little linksys trash would be using it too. The enterprise hardware isn't as price sensitive as the consumer/retail market, so they can afford to go a bit overboard with the hardware.
They went back to VxWorks. Their claim is it's smaller, so they can put smaller (cheaper) flash chips, and less RAM in the systems. But I seriously doubt the cost savings amounts to what they have to pay Wind River in licensing.
Both OpenWRT and DD-WRT produce fully opensource firmwares for a number of systems, including the crippled 54g v5+. All of my linksys toys run dd-wrt (well, except the uber-ancient WAP11, but it's just an atmel microcontroller with a PCMCIA wireless card bolted to it.)