I read through the whole article, and they make some very good points. I notice a tendency of Linux fanatics to not read articles and understand what the author is saying before jumping to flame them.
And their summary really put everything into perspective:
"There is nothing strange or magical about open-source development from a business point of view; it should neither be shunned as impractical nor embraced as a panacea," Mr. Hecker concludes in his white paper.
Each generation of CPU Intel has come out with has been just a whee bit better and first, and then suddenly it became a whole lot better.
Going from the 386DX33 to 486DX25 was no big deal. Going from a 486DX2/66 to a P-60 was no big deal, in fact the P60 overheated so much people generally didn't recommend it. By the time the P75/P90 came out we had 486DX4/100 which was only slightly slower.
It wasn't until the P200 came out that things really sped up. And then the first PII was a 233 Mhz, which was quite a bit faster, but still not as awe inspiring as the PII-450.
The Celeron 233 and 300 at first were called flops, then the 300A and 333A came out and suddenly their fantastic CPUs for a low price?
So the PIII-500 isn't much faster... Intel's anticipating 800 Mhz available later this year.
It'll sell. Maybe not the first version, but later revisions will become the standard, as always.
But first we need to find the CPU bug and fix it. Don't know what the bug is, but the chip has got to have one in there somewhere, which of course will generate all sorts of media hype!:)
Actually the big thing right now is frankly that the system bus, memory, harddrives and other I/O cannot keep up with the CPU.
You have to remember these benchmarks like WinBench 99 are measuring complete system performance. And frankly to be honest, most people don't need a PII-450 much less a PIII-800. However there are some applications out there which do heavy compuational analysis, and for them a PIII may very well be quite a big thing.
How can one possibly write a paper on the history of computing/programming and make absolutely no mention of Iowa State University where the first electronic digital computer was invented?
But then Mauchly and Eckert were working in the great Open Source tradition of taking Atanasoff's work, building upon it, and getting all the credit.:)
No mention of Hollerith, or Babbage either. And what about Blaise Pascal and...:)
I'd be particularly curious to see the same tests run on a tuned environment.
Say a Proliant 6500, Quad processor, 512 Meg RAM, 100 gigs on RAID-5, etc...
That's going to be a fairly standard corporate server, and it'd be curious to see the results with say 500 client computers.
Unfortunately only a few places are going to have the hardware to do such benchmarking. I understand that Compaq has such a lab available.
Also I don't believe there was mention of what kind of netbench test that was, it must have just been static content? What about dynamic content from asp or php or cgi or whatever?
I'd also be curious how one might try to benchmark Samba as a print server. Most companies don't just install servers for file sharing, but also for printer sharing...
I think the use of harddrives for data storage is going to cause them problems, as they will be unable to withstand the environment presented by an automobile.(extreme heat and cold, as well as shock)
I'm sorry, but I have to agree with the RIAA on this. If the guy is taking this copyrighted material putting it on a site with advertisements generating revenue, then he needs to be working with the RIAA on compensating the artists for their work.
I think it's amazing how many people are complaining about RIAA greed, yet seem perfectly happy with someone stealing others intellectual property and making money off it. That's not just greed, that's being a leech.
If it's true that he makes no money, then it shouldn't be too hard to make his accounting information public detailing said claim.
I think you'll find that most artists and record labels don't have a problem with having their lyrics available. But like the article said, they do have a problem with people taking that work and making money off it without compensation.
Right off the bat one of ESR's claims is rather dubious. He claims that all the Internet core software is open source and subject to peer evaluation. Where it is not, it falls down and fails.
That's rather naive. A large number of the major players are not open source, including most of the commerce web servers built upon Netscape, etc. Along with web search engines, email systems like hotmail, etc.
But more important, I was not aware that Cisco and the other router vendors had made their internal software open source.
And it's the routers and other data communications equipment which is the core of the internet.
Some interesting points made, but I certainly don't think he's proven any of them.
I think you misunderstand the source code requirement. I suspect having the source code easily available to the end-user, i.e. Opensource, would tend to violate the intent of the requirement, even if it is not installed.
I do some charity work with the Jaycee's and other organizations.
In some cases you can get a tax deduction for donating money, or mileage on your automobile.
But I'm not aware of any charitable donation tax write-off for your time.
There is also already a pretty strong impression that OpenSource "Free" software is not really free, but paid for by tax dollars. Do you really want to continue this negative impression?
I'd be curious to see how GRASS has changed. I last used it in 1995 and it really pretty much sucked at the time. We did however use LTPlus for our map generation and it was a slick program.
Arc/Info was bloated, it was expensive. But damn did it do everything, and it worked pretty well. Because the application was so massive, ESRI did tend to release buggy code, and you really wanted to wait for like version x.1 or x.2 before using it.
Arc/Info really wasn't a end-user package. Everything you did was pretty much done by writing macros. ArcView was their end-user package, although they also bought Strategic Mapping out a couple years ago so now have Atlas*GIS.
Ohwell, I miss Arc/Info and my mapping days to some extent.:(
I don't know much about mappoint, but the mapping package which is integrated with Microsoft Office was built by Strategic Mapping, makers of Atlas*GIS.
ESRI bought them out about two years ago. Perhaps Microsoft bought the rights to that product during the ESRI acquisition?
Perhaps per minute is a bit of a stretch. Still just one server on 100baseT could server about 9 Meg/sec or rougly 9*60 = 540 Meg/minute. That's still substantially faster than the stupid 6 Gig/day claim.
As far as cost. Even if you do consider 3 servers with $2100 in software costs...
You still aren't going to be able to buy a server for that little. A good sized Proliant is going to set you back $20k, probably more like $40k when you throw in UPS, DLT drives and such. Multiply that by three as you apparantely desire to do.
As far as multiple machines... Consider you also need machines to handle peak capacity. Or do you think sites like CNN.COM and YAHOO.COM are only run off one machine?
Oh, and machines die from more than just harddrives.
Let's see we're talking about a large FTP site here:
"and the software license will cost about as much as the server"
What the hell kind of server can you buy for $700? That's all the NT Server license will cost you.
"you can expect to move about 6 gigabytes (GB) a day from three Compaq ProLiant 5000s "
That's not what the Microsoft article says. It mentions that it handles an average of 6 gigs of downloads a day.
Microsoft doesn't run a large download server such as cdrom.com. There primary content is web pages, with a few downloads. Obviously when something new is released that requires a download such as NT Service Pack, DirectX version, etc. they get hit hard and have to have server capacity in place to accomodate that increased load.
NT4 SP4 was about a 75 meg download with all the extra stuff. 6 Gigs a day would allow maybe 70 people to download it, which is more likely the number of people who downloaded it every minute.
Let's see, what else:
"In order to maintain availability, Microsoft recommends that you install multiple systems with failover. "
Doesn't everybody?
If you have a hardware failure, can you afford for your site to be down several hours or even several days while you fix it?
I used to subscribe to SunWorld years ago, and then they stopped their paper publication and went to this web based magazine. I haven't bothered to read it for the last year or so, and it appears that the quality of their articles has suffered tremendously.
I can only assume a moron wrote this article. I'll have to send him email as well to help educate him.
4.Infiltrate the newsgroups (SlashDot et. al) with inflammatory comments making more rift in the Linux camps and then spread more FUD on fragmented the so-called "Linux-Community" is.
Heh. The Linux community does quite well at this by themselves. It started with Slackware and the guy was going to *shock* charge for a CD. And it's extended itself right on into the GNOME/KDE wars, and everything else...
I don't think people understand why Intel is so against Overclocking.
It isn't because some poor schmuck can't afford a 400 Mhz chip, so he buys a 350 Mhz chip.
It's because some poor schmuck who can afford to buy a 400 Mhz chip walks into CULater computers plunks down the cash to buy what he thinks is a machine with a 400 Mhz chip in it, only to actually get a 300 Mhz chip overclocked.
Yeah, sure it might work. But it's dishonest. The customer essentially gets screwed, paying more for less. Intel certainly cares because they had a customer willing to pay for the more expensive chip, but instead the money went to line the pocket of the dishonest dealer.
I had a company pull that on me once. It was quite a few years ago, but a machine came overclocked and we had tons of trouble with it. That was when I discovered the machine was mislabeled and boy did I raise holy hell with the vendor.
Slashdot Mirror
I read through the whole article, and they make some very good points. I notice a tendency of Linux fanatics to not read articles and understand what the author is saying before jumping to flame them.
And their summary really put everything into perspective:
"There is nothing strange or magical about open-source development from a business point of view; it should neither be shunned as impractical nor embraced as a panacea," Mr. Hecker concludes in his white paper.
Each generation of CPU Intel has come out with has been just a whee bit better and first, and then suddenly it became a whole lot better.
:)
Going from the 386DX33 to 486DX25 was no big deal. Going from a 486DX2/66 to a P-60 was no big deal, in fact the P60 overheated so much people generally didn't recommend it. By the time the P75/P90 came out we had 486DX4/100 which was only slightly slower.
It wasn't until the P200 came out that things really sped up. And then the first PII was a 233 Mhz, which was quite a bit faster, but still not as awe inspiring as the PII-450.
The Celeron 233 and 300 at first were called flops, then the 300A and 333A came out and suddenly their fantastic CPUs for a low price?
So the PIII-500 isn't much faster... Intel's anticipating 800 Mhz available later this year.
It'll sell. Maybe not the first version, but later revisions will become the standard, as always.
But first we need to find the CPU bug and fix it. Don't know what the bug is, but the chip has got to have one in there somewhere, which of course will generate all sorts of media hype!
Actually the big thing right now is frankly that the system bus, memory, harddrives and other I/O cannot keep up with the CPU.
You have to remember these benchmarks like WinBench 99 are measuring complete system performance. And frankly to be honest, most people don't need a PII-450 much less a PIII-800. However there are some applications out there which do heavy compuational analysis, and for them a PIII may very well be quite a big thing.
"The ones I saw look pretty smooth. "
Were you talking about the hats, or the heads?
They point out in the article that 10base2 uses less power than 10baseT.
That's kind of important when you're running everything off large solar cells and batteries.
Not to mention a 10base2 LAN will require less cable than a 10baseT LAN, as well as a hub. This saves a few ounces of weight...
This is important when you are flying around 250 miles above the planet.
How can one possibly write a paper on the history of computing/programming and make absolutely no mention of Iowa State University where the first electronic digital computer was invented?
:)
:)
But then Mauchly and Eckert were working in the great Open Source tradition of taking Atanasoff's work, building upon it, and getting all the credit.
No mention of Hollerith, or Babbage either. And what about Blaise Pascal and...
I'd be particularly curious to see the same tests run on a tuned environment.
Say a Proliant 6500, Quad processor, 512 Meg RAM, 100 gigs on RAID-5, etc...
That's going to be a fairly standard corporate server, and it'd be curious to see the results with say 500 client computers.
Unfortunately only a few places are going to have the hardware to do such benchmarking. I understand that Compaq has such a lab available.
Also I don't believe there was mention of what kind of netbench test that was, it must have just been static content? What about dynamic content from asp or php or cgi or whatever?
I'd also be curious how one might try to benchmark Samba as a print server. Most companies don't just install servers for file sharing, but also for printer sharing...
I think the point is still that the design is not really suitable for the environment it will exist in.
That's poor engineering, frankly.
Again I think it's a neat idea, but it seems to be intended as one persons pet project rather than a commercially viable device.
Yup,
After reading the description of the patent I immediately realized just how much of this has been implemented in Netrek.
Not recently either, this has been around for 10 years or more.
Pathetic... I'm going to send them a letter along with one to the patent office.
Steve
Netrek Client/Server coding geek
Looks like a neat idea.
I think the use of harddrives for data storage is going to cause them problems, as they will be unable to withstand the environment presented by an automobile.(extreme heat and cold, as well as shock)
I'm sorry, but I have to agree with the RIAA on this. If the guy is taking this copyrighted material putting it on a site with advertisements generating revenue, then he needs to be working with the RIAA on compensating the artists for their work.
I think it's amazing how many people are complaining about RIAA greed, yet seem perfectly happy with someone stealing others intellectual property and making money off it. That's not just greed, that's being a leech.
If it's true that he makes no money, then it shouldn't be too hard to make his accounting information public detailing said claim.
I think you'll find that most artists and record labels don't have a problem with having their lyrics available. But like the article said, they do have a problem with people taking that work and making money off it without compensation.
Right off the bat one of ESR's claims is rather dubious. He claims that all the Internet core software is open source and subject to peer evaluation. Where it is not, it falls down and fails.
That's rather naive. A large number of the major players are not open source, including most of the commerce web servers built upon Netscape, etc. Along with web search engines, email systems like hotmail, etc.
But more important, I was not aware that Cisco and the other router vendors had made their internal software open source.
And it's the routers and other data communications equipment which is the core of the internet.
Some interesting points made, but I certainly don't think he's proven any of them.
The URL I posted is correct.
I think you misunderstand the source code requirement. I suspect having the source code easily available to the end-user, i.e. Opensource, would tend to violate the intent of the requirement, even if it is not installed.
I do some charity work with the Jaycee's and other organizations.
In some cases you can get a tax deduction for donating money, or mileage on your automobile.
But I'm not aware of any charitable donation tax write-off for your time.
There is also already a pretty strong impression that OpenSource "Free" software is not really free, but paid for by tax dollars. Do you really want to continue this negative impression?
http://www.itl.nist.gov/div897/pubs/fip140-1.htm
:-)
Neat, Linux also fails FIPS 140-1 since it does not meet a requirement of Security Level 1 for OS design:
"- All cryptographic software shall be installed only as executable code in order to discourage scrutiny and modification by users. "
Heh
Steve
I'd be curious to see how GRASS has changed. I last used it in 1995 and it really pretty much sucked at the time. We did however use LTPlus for our map generation and it was a slick program.
:(
Arc/Info was bloated, it was expensive. But damn did it do everything, and it worked pretty well. Because the application was so massive, ESRI did tend to release buggy code, and you really wanted to wait for like version x.1 or x.2 before using it.
Arc/Info really wasn't a end-user package. Everything you did was pretty much done by writing macros. ArcView was their end-user package, although they also bought Strategic Mapping out a couple years ago so now have Atlas*GIS.
Ohwell, I miss Arc/Info and my mapping days to some extent.
Steve
I don't know much about mappoint, but the mapping package which is integrated with Microsoft Office was built by Strategic Mapping, makers of Atlas*GIS.
ESRI bought them out about two years ago. Perhaps Microsoft bought the rights to that product during the ESRI acquisition?
Perhaps per minute is a bit of a stretch. Still just one server on 100baseT could server about 9 Meg/sec or rougly 9*60 = 540 Meg/minute. That's still substantially faster than the stupid 6 Gig/day claim.
As far as cost. Even if you do consider 3 servers with $2100 in software costs...
You still aren't going to be able to buy a server for that little. A good sized Proliant is going to set you back $20k, probably more like $40k when you throw in UPS, DLT drives and such. Multiply that by three as you apparantely desire to do.
As far as multiple machines... Consider you also need machines to handle peak capacity. Or do you think sites like CNN.COM and YAHOO.COM are only run off one machine?
Oh, and machines die from more than just harddrives.
Let's see we're talking about a large FTP site here:
"and the software license will cost about as much as the server"
What the hell kind of server can you buy for $700? That's all the NT Server license will cost you.
"you can expect to move about 6 gigabytes (GB) a day from three Compaq ProLiant 5000s "
That's not what the Microsoft article says. It mentions that it handles an average of 6 gigs of downloads a day.
Microsoft doesn't run a large download server such as cdrom.com. There primary content is web pages, with a few downloads. Obviously when something new is released that requires a download such as NT Service Pack, DirectX version, etc. they get hit hard and have to have server capacity in place to accomodate that increased load.
NT4 SP4 was about a 75 meg download with all the extra stuff. 6 Gigs a day would allow maybe 70 people to download it, which is more likely the number of people who downloaded it every minute.
Let's see, what else:
"In order to maintain availability, Microsoft recommends that you install multiple systems with failover. "
Doesn't everybody?
If you have a hardware failure, can you afford for your site to be down several hours or even several days while you fix it?
I used to subscribe to SunWorld years ago, and then they stopped their paper publication and went to this web based magazine. I haven't bothered to read it for the last year or so, and it appears that the quality of their articles has suffered tremendously.
I can only assume a moron wrote this article. I'll have to send him email as well to help educate him.
I like this:
4.Infiltrate the newsgroups (SlashDot et. al) with inflammatory comments making more rift in the Linux camps and then spread more FUD on fragmented the so-called "Linux-Community" is.
Heh. The Linux community does quite well at this by themselves. It started with Slackware and the guy was going to *shock* charge for a CD. And it's extended itself right on into the GNOME/KDE wars, and everything else...
Yes the AT command set worked brilliantly, and it gave a standard method of communicating with the modem.
Other manufacturers agreed and implemented it in their modems, making it incredibly easy to write terminal programs and such back in the early 80's.
That's about when Hayes decided to sue everybody who was using the AT command set for copyright infringement.
In a way Hayes sort of killed themselves. Their last good product was the Hayes Smartmodem 1200.
(sarcasm on)
Quit trolling you obvious Microsoft lover!
(sarcasm off)
I don't think people understand why Intel is so against Overclocking.
It isn't because some poor schmuck can't afford a 400 Mhz chip, so he buys a 350 Mhz chip.
It's because some poor schmuck who can afford to buy a 400 Mhz chip walks into CULater computers plunks down the cash to buy what he thinks is a machine with a 400 Mhz chip in it, only to actually get a 300 Mhz chip overclocked.
Yeah, sure it might work. But it's dishonest. The customer essentially gets screwed, paying more for less. Intel certainly cares because they had a customer willing to pay for the more expensive chip, but instead the money went to line the pocket of the dishonest dealer.
I had a company pull that on me once. It was quite a few years ago, but a machine came overclocked and we had tons of trouble with it. That was when I discovered the machine was mislabeled and boy did I raise holy hell with the vendor.