If you read the areticle carefully, you'd find that the bandwidth has been pushed up to allow a 1920x1080x24bit (HDTV 1080) display to be update at 90hz. That would allow a 2048×1536 (think Apple 30" Cinema Display) to be run at 60hz with a Type A connector. That's an interesting development.
It doesn't in a different league from current HD. I don't see them with a different value proposition from the traditional competitors and yet they have this completely new and untested technology. I'm afraid they would go too soon on the market.
Just try OpenBSD with it's pf, carp, pfsync, bgpd, ipsecctl, ospfd and all systraced. You can get a fault tolerant, load balancing encryption accelerating rig with two Via ITX boxes (somehting like $900 in overall cost).
RTFA!!! OpenBSD wants only free distribution right of the binary files. That's all, they don't even want the right to modify the binary. Just to be able to distribute it like they do with so many other firmware files. Else you have to go download the file from somewhere else (how do you do that without a network connection) or OpenBSD has to sign an agreement that they won't since they would have to limit the way they distribute their software. Among other things, they would have to put you through a click through license when installing OpenBSD!!!!
So emerge -uK is insulting to Engligh and thus should not be used? I use and swear by OpenBSD, but you can't really include Gentoo and a critique of lack of pgk_get for Linux in a same paragraph.
Whre I live you can get Ar $1.5 per hour. And that's in the main city. Since exchange rate is Ar$3=U$S1 it's like 50 US cents per hour. The only way they could reach that cost is on certain cibercafes on remote tourist zones where they might cost that. I've only seen two and those where on places where only hicking and alpinist tourists go. The overwhelming amount of population has U$S0.66 an hour internet in this country. So I might take a serious dubt about the veracity of those numbers.
4) Default security levels. For me it's crucial to have a healthy server. Let's take OpenBSD, you have under 10 processes in the default install. And practically any dangerous setting is turned off. It's really effort less to install stuff, just turn on what you want. If the default install is not enough for you, then it might have some security concerns. It's so easy to start from a blank page. Ditto for Gentoo, but OpenBSD installs in just 10 minutes, including partitioning and network configuration. I don't know about Slackware, tough. Now, compare that to RH 7.2 (which was the last version I've installed). I stopped using it because it took me 1h 30m to get an install without adding XFree86! Never mind the amount of unsafe processes running by default. Mandrake was a bit better but was so unprofessional and full of bugs that I've given up. Ok, this might seem like a biased view of a server admin, but the truth is that security is being _the_ hot topic this last few years.
In which sense? Don't we truly think that the GPL is valid? Lousy lawyers are invalid in court? I never ever said that he had to tell the lawyer to loose. He simply won't. Does anyone with more knowledge of law can give an opinion?
What I still don't undestand is why some true geek doesn't purposedly inflinges the GPL, is sent to court and hires the lousiest lawyer he can find. Repeat a hunded times (since the lawyer is so bad you can have them cheaply). Now you have a greatly tested in court Licence! Why some rich dotcommer doesn't does this a contribution to the community?
First are a standard. And second the do solve a problem. Namely, the fact that the maximum bandwidth is achievable is the packet size _times_ the inverse of the round trip latency. So if you have a big round trip latency, you can't achieve high bandwidth, at least with a single tranfer.
No network geek would ever use Cisco! If we play the "money is no problem" game, then go for a Juniper. At least you get good native IPv6 support and VLAN done right. Most switches (specially Cisco) are a real disgrace for security. You can easily overflow the MAC buffers, in which case they will act as hubs (so you can kiss your VLANs goodbye.) Not to mention the common exploits to IOS. And the fact the Cisco plays the patents game to not let anyone else have VRRP. And that they are outrageously expensive. And if you want to talk about security, don't even mention a PIX the favourite piece for a script kiddy. Just buy a couple of Tyan Tiger GC-SL with integrated 1Gbps ports motherboards, put just _one_ cpu on each. Put three Sysconnect SK-9822 (those are dual ports server cards) on each. Now install OpenBSD 3.5 on each and you have a fault tolerant, 7 ports, Gigabit firewall solution for much, much less than an equivalent PIX. Plus you have enough ports not to need VLAN when segmenting your network.
Well,
Buenos Aires has a 40% of Spanish, 25% Italian, 9% Lebanese (which are the descendants of the crusaders and Christians), 8% Jews, 7% of other European (mostly German, English, Polish and Russian) and just an 11% of natives grandfathers.
So if you where to pick anyone at random in buenos Aires chances are great that you won't pick what Americans call Latinos, which is what we call mestizo. If you consider Latinos as having a Latin-rooted mother tongue (which includes Italians, Portuguese, Spanish, Frenchs and Rumanians) then, yes you will mostly find a latin population.
Again: 64 bits is enough to address every single molecule in Earth. So unless you get into subatom memory cells, 128bit won't be needed for addressing. Regarding operands, may be. But going that way will be very difficult (PowerPC is already there with FP, thou).
That means that some clienta has made a huge investment in Linux/IBM. So, should SCO win the case, what they loose from that deal they make up by owning 20% of SCO. Or it's simply a M$ front end.
That is if the best cracking method is brute force search. Which for AES and such should be (but might be less, we don't yet know). But for RSA it's definitelly not. What you have to do is to factorize the two primes embedded in the public key to know the private key. It's known that the best algorithms for doing that are much better than o(2^n). But what really helps is bignumber support (for factoring), where this DSP is slow, and fast matrix inversion (for solving the linear equations), where this DSP is fast. But it's my understanding that even i you could achieve inmediate matrix inverions, you'd still have todo the factoring. If with current technology you have a 50% of time for each step (I don't realy know the proportion). Having instantaneous matrix inversion (an improvemen of NaN%:-) only represents a 100% improvement in the execution time.
I can't remember the article. But in Canada they put some data to go rounds through the national backbone. It was an experiment tu use excess bandwidth as storage cache. But it must have broken the distance record by a wide margin.
I mean, P0rn DVD are just 70 to 80 minutes. And I can't think of anything else for so much bandwidth. LOTR trilogy extended edition might do, but just for the bragging rights. No... p0rn should do...:-)
I had a similar path. I use mostly Windows workstations (I'm an Elder Scrolls addict:-). But mostly with OoO, Mozilla, The Bat!, and gretl (econometrics, never mind). But on the servers I use exclusively Linux. I started with a Slackware with a 1.3 development kernel. Thou I didn't really got into Linux until RH 5.0. I'd been Netware guy before. Anyway. Around 2000 I bought an OpenBSD 2.7 for a firewall (it used ipf by that time). And inmediately fall in love with it. I't an amazing firewal/router. And over the years I've come to understand that what seems like a spartan install it's actually the only OS I can install in under 5 minutes. Besides I can have an aditional site00.tgz with all my config files so if I burn the CD myself I can have the server up and running in 5 minutes. Just amazing. Since I use it only for router/firewall (dhcp, ntp, dns, vpn and all this little services) I've never installed X. But I've found a couple of things. Once you understan to use the man page and the FAQ it's the easiest, simplest system around. Do you have a dubt about how to configure anything? use man. Have a dubt about a system interace? use man. Have a dubt about the syntax of a command? use man. Compare this to the HOW-TO, man, info, hadnbook, mailing list, googling of any Linux instalation. And the other thins that sold me is the IPv6 support. Everything works with it. You can have the reference IKA implementation (see WIDE.org) running on IPv6. I couldn't believe when I bought my XP laptop that I was actually seeing the KAME turtle swimming! And this was thanks to my router that had hidden all the uglies of IPv6 configuration. It's even hardware accelerated (Linux, NetBSD and FreeBSD guys had to import the crypto framework from OpenBSD). I can't really not mention the true jewel of OpenBSD, pf. It simply has, quite simply, the most awesome team of developers on IP filtering around. I was used to the Linux world when someone came with a stateless ip firewall. Then someone else came with a NAT extension. Then they rewrote the filter with the nat extension. Then there was a couple of projects with quality of service. Then they rewrote the whole thing to make it pseudostateful. Then, some patches crept out for some foo. And so you want to hide the amount of PC you're NATing? Find the patch and hope it's compatible. Want to use strong sequence numbers for all broken client's? Hope there's a patch somewhere. With pf you have twice a year the guy telling the new features they integrate for the next OS version. They went form making it a 80% of pf in three monthes, to making it 120% in the next six. Then they integrated accounting, bandwidth and pririty queuing, tagging at the ethernet level, IPSec level, dinamic inclusion and deletion of rules as well as addresses on tables. And so many things I can't even remember. And all the meanwhile you have the little treats like the week next to the paper explaining how to count the number of clients behind NAT, thay integrated a new keword that defeated it. Ditto for the weak sequence numbers. Ditto for the spam tarpit. And the unbelivable thing is: I could use the same configuration file from the very first pf version. And mostly form any ipf version. I had to rewrite it to take advantage of new features, but that's expected. So I'm absolutely sold to OpenBSD on the router. For the server I'm still a Gentoo guy. I might try a bit of FreeBSD or RedHat when they get their act together with AMD64. But for now it's kind of the perfect combination.
Great! Now tell me. People will just remember their IPs? They won't check their POP/IMAP boxes? Won't play CounterStrike or whatever is cool game of the moment? Think a bit about what you just said. If companies don't want their employees to use anything _but_ HTTP/S because they should use the internal mail system and DNS cache and such stuff. It's OK. That's why they are companies. But ISP should not filter anything. Besides, will they start to block AH or ESP? What about 6to4? And ICMP and ARP? I don't think so.
I've found that the problem with most statistical tests is the H0 (i.e. the question we are asking about some parameter). Don't confuse with the questions you may ask in the survey but rather what you are wondering about the market. E.g. if you are asking what were the purchased OS of the Fortune 500 companies is different to the actual OS deployed throught anybody with a non-personal server connected to the internet. Where do you state that information clearly and how do you design the survey for a given question?
How do you design your sampling methods when such questions are answered? How do you deside which errors are you more willing to accept? Because you may design a low power question if it's less biased than other or are you willing to sacrifice a bit more of Type I/II error for a less biased survey desing? Who makes this decisions and with which rules?
Do you test the sample for statistical properties? Do you segment your sample candidates to represent your desired universe?
I understand the funny thing about the article. But, does Gateway had any PR problems which might have inspired this satiric? Because I do remember that IBM had some problems with their TP (I own 2 and I'm looking for my third, thou).
Don't forget that Argentina is so scarcely populated that we don't have a railway to the South. And Tierra del Fuego is an island, so you'd need more tunnels (keep the zeros goings!).
BTW, if the cost of it were like U$S 250M I'm sure our gov would quote U$S 2000M (8:1 rule of corruption). But I've heard that Russian is still more "inefficient" (I would guess it's 40:1, but I don't have enough experience with that gov).
Well, I wouldn't call the i740 bleeding edge, the i810 a price/performance champion, the i820 or i840 a robust platform. Recall the how late the P!!! and 820 were (reliability does includes promised launch dates). Consider how the crippled the Celeron by keeping its FSB at 66Mhz. Recall how they had to cede the whole server/workstation chipset market to ServerWorks just because they had to use RDRAM.
No, I'm not the typical slashdotter and yet I've lost all my respecto for Intel.
Slashdot Mirror
Proctor & Gamble? I think P&G will be over those perverts with a brandname suit in no time. :p
If you read the areticle carefully, you'd find that the bandwidth has been pushed up to allow a 1920x1080x24bit (HDTV 1080) display to be update at 90hz. That would allow a 2048×1536 (think Apple 30" Cinema Display) to be run at 60hz with a Type A connector. That's an interesting development.
It doesn't in a different league from current HD. I don't see them with a different value proposition from the traditional competitors and yet they have this completely new and untested technology. I'm afraid they would go too soon on the market.
Just try OpenBSD with it's pf, carp, pfsync, bgpd, ipsecctl, ospfd and all systraced. You can get a fault tolerant, load balancing encryption accelerating rig with two Via ITX boxes (somehting like $900 in overall cost).
RTFA!!!
OpenBSD wants only free distribution right of the binary files. That's all, they don't even want the right to modify the binary. Just to be able to distribute it like they do with so many other firmware files. Else you have to go download the file from somewhere else (how do you do that without a network connection) or OpenBSD has to sign an agreement that they won't since they would have to limit the way they distribute their software. Among other things, they would have to put you through a click through license when installing OpenBSD!!!!
So emerge -uK is insulting to Engligh and thus should not be used?
I use and swear by OpenBSD, but you can't really include Gentoo and a critique of lack of pgk_get for Linux in a same paragraph.
Whre I live you can get Ar $1.5 per hour. And that's in the main city. Since exchange rate is Ar$3=U$S1 it's like 50 US cents per hour. The only way they could reach that cost is on certain cibercafes on remote tourist zones where they might cost that. I've only seen two and those where on places where only hicking and alpinist tourists go.
The overwhelming amount of population has U$S0.66 an hour internet in this country. So I might take a serious dubt about the veracity of those numbers.
4) Default security levels.
For me it's crucial to have a healthy server. Let's take OpenBSD, you have under 10 processes in the default install. And practically any dangerous setting is turned off. It's really effort less to install stuff, just turn on what you want. If the default install is not enough for you, then it might have some security concerns. It's so easy to start from a blank page. Ditto for Gentoo, but OpenBSD installs in just 10 minutes, including partitioning and network configuration. I don't know about Slackware, tough.
Now, compare that to RH 7.2 (which was the last version I've installed). I stopped using it because it took me 1h 30m to get an install without adding XFree86! Never mind the amount of unsafe processes running by default. Mandrake was a bit better but was so unprofessional and full of bugs that I've given up.
Ok, this might seem like a biased view of a server admin, but the truth is that security is being _the_ hot topic this last few years.
In which sense? Don't we truly think that the GPL is valid? Lousy lawyers are invalid in court? I never ever said that he had to tell the lawyer to loose. He simply won't. Does anyone with more knowledge of law can give an opinion?
What I still don't undestand is why some true geek doesn't purposedly inflinges the GPL, is sent to court and hires the lousiest lawyer he can find. Repeat a hunded times (since the lawyer is so bad you can have them cheaply).
Now you have a greatly tested in court Licence!
Why some rich dotcommer doesn't does this a contribution to the community?
First are a standard. And second the do solve a problem. Namely, the fact that the maximum bandwidth is achievable is the packet size _times_ the inverse of the round trip latency. So if you have a big round trip latency, you can't achieve high bandwidth, at least with a single tranfer.
No network geek would ever use Cisco!
If we play the "money is no problem" game, then go for a Juniper. At least you get good native IPv6 support and VLAN done right. Most switches (specially Cisco) are a real disgrace for security. You can easily overflow the MAC buffers, in which case they will act as hubs (so you can kiss your VLANs goodbye.) Not to mention the common exploits to IOS. And the fact the Cisco plays the patents game to not let anyone else have VRRP. And that they are outrageously expensive.
And if you want to talk about security, don't even mention a PIX the favourite piece for a script kiddy. Just buy a couple of Tyan Tiger GC-SL with integrated 1Gbps ports motherboards, put just _one_ cpu on each. Put three Sysconnect SK-9822 (those are dual ports server cards) on each. Now install OpenBSD 3.5 on each and you have a fault tolerant, 7 ports, Gigabit firewall solution for much, much less than an equivalent PIX. Plus you have enough ports not to need VLAN when segmenting your network.
Well,
Buenos Aires has a 40% of Spanish, 25% Italian, 9% Lebanese (which are the descendants of the crusaders and Christians), 8% Jews, 7% of other European (mostly German, English, Polish and Russian) and just an 11% of natives grandfathers.
So if you where to pick anyone at random in buenos Aires chances are great that you won't pick what Americans call Latinos, which is what we call mestizo. If you consider Latinos as having a Latin-rooted mother tongue (which includes Italians, Portuguese, Spanish, Frenchs and Rumanians) then, yes you will mostly find a latin population.
Again: 64 bits is enough to address every single molecule in Earth. So unless you get into subatom memory cells, 128bit won't be needed for addressing. Regarding operands, may be. But going that way will be very difficult (PowerPC is already there with FP, thou).
That means that some clienta has made a huge investment in Linux/IBM. So, should SCO win the case, what they loose from that deal they make up by owning 20% of SCO.
Or it's simply a M$ front end.
That is if the best cracking method is brute force search. Which for AES and such should be (but might be less, we don't yet know). But for RSA it's definitelly not. What you have to do is to factorize the two primes embedded in the public key to know the private key. It's known that the best algorithms for doing that are much better than o(2^n). :-) only represents a 100% improvement in the execution time.
But what really helps is bignumber support (for factoring), where this DSP is slow, and fast matrix inversion (for solving the linear equations), where this DSP is fast. But it's my understanding that even i you could achieve inmediate matrix inverions, you'd still have todo the factoring. If with current technology you have a 50% of time for each step (I don't realy know the proportion). Having instantaneous matrix inversion (an improvemen of NaN%
I can't remember the article. But in Canada they put some data to go rounds through the national backbone. It was an experiment tu use excess bandwidth as storage cache. But it must have broken the distance record by a wide margin.
I mean, P0rn DVD are just 70 to 80 minutes. And I can't think of anything else for so much bandwidth. LOTR trilogy extended edition might do, but just for the bragging rights. No... p0rn should do... :-)
I had a similar path. I use mostly Windows workstations (I'm an Elder Scrolls addict :-). But mostly with OoO, Mozilla, The Bat!, and gretl (econometrics, never mind).
But on the servers I use exclusively Linux. I started with a Slackware with a 1.3 development kernel. Thou I didn't really got into Linux until RH 5.0. I'd been Netware guy before. Anyway. Around 2000 I bought an OpenBSD 2.7 for a firewall (it used ipf by that time). And inmediately fall in love with it. I't an amazing firewal/router. And over the years I've come to understand that what seems like a spartan install it's actually the only OS I can install in under 5 minutes. Besides I can have an aditional site00.tgz with all my config files so if I burn the CD myself I can have the server up and running in 5 minutes. Just amazing.
Since I use it only for router/firewall (dhcp, ntp, dns, vpn and all this little services) I've never installed X. But I've found a couple of things. Once you understan to use the man page and the FAQ it's the easiest, simplest system around. Do you have a dubt about how to configure anything? use man. Have a dubt about a system interace? use man. Have a dubt about the syntax of a command? use man. Compare this to the HOW-TO, man, info, hadnbook, mailing list, googling of any Linux instalation.
And the other thins that sold me is the IPv6 support. Everything works with it. You can have the reference IKA implementation (see WIDE.org) running on IPv6. I couldn't believe when I bought my XP laptop that I was actually seeing the KAME turtle swimming! And this was thanks to my router that had hidden all the uglies of IPv6 configuration. It's even hardware accelerated (Linux, NetBSD and FreeBSD guys had to import the crypto framework from OpenBSD).
I can't really not mention the true jewel of OpenBSD, pf. It simply has, quite simply, the most awesome team of developers on IP filtering around. I was used to the Linux world when someone came with a stateless ip firewall. Then someone else came with a NAT extension. Then they rewrote the filter with the nat extension. Then there was a couple of projects with quality of service. Then they rewrote the whole thing to make it pseudostateful. Then, some patches crept out for some foo. And so you want to hide the amount of PC you're NATing? Find the patch and hope it's compatible. Want to use strong sequence numbers for all broken client's? Hope there's a patch somewhere.
With pf you have twice a year the guy telling the new features they integrate for the next OS version. They went form making it a 80% of pf in three monthes, to making it 120% in the next six. Then they integrated accounting, bandwidth and pririty queuing, tagging at the ethernet level, IPSec level, dinamic inclusion and deletion of rules as well as addresses on tables. And so many things I can't even remember.
And all the meanwhile you have the little treats like the week next to the paper explaining how to count the number of clients behind NAT, thay integrated a new keword that defeated it. Ditto for the weak sequence numbers. Ditto for the spam tarpit. And the unbelivable thing is: I could use the same configuration file from the very first pf version. And mostly form any ipf version. I had to rewrite it to take advantage of new features, but that's expected.
So I'm absolutely sold to OpenBSD on the router. For the server I'm still a Gentoo guy. I might try a bit of FreeBSD or RedHat when they get their act together with AMD64. But for now it's kind of the perfect combination.
Great! Now tell me. People will just remember their IPs? They won't check their POP/IMAP boxes? Won't play CounterStrike or whatever is cool game of the moment? Think a bit about what you just said. If companies don't want their employees to use anything _but_ HTTP/S because they should use the internal mail system and DNS cache and such stuff. It's OK. That's why they are companies. But ISP should not filter anything. Besides, will they start to block AH or ESP? What about 6to4? And ICMP and ARP? I don't think so.
I've found that the problem with most statistical tests is the H0 (i.e. the question we are asking about some parameter). Don't confuse with the questions you may ask in the survey but rather what you are wondering about the market. E.g. if you are asking what were the purchased OS of the Fortune 500 companies is different to the actual OS deployed throught anybody with a non-personal server connected to the internet. Where do you state that information clearly and how do you design the survey for a given question?
How do you design your sampling methods when such questions are answered?
How do you deside which errors are you more willing to accept? Because you may design a low power question if it's less biased than other or are you willing to sacrifice a bit more of Type I/II error for a less biased survey desing? Who makes this decisions and with which rules?
Do you test the sample for statistical properties? Do you segment your sample candidates to represent your desired universe?
I understand the funny thing about the article. But, does Gateway had any PR problems which might have inspired this satiric? Because I do remember that IBM had some problems with their TP (I own 2 and I'm looking for my third, thou).
Don't forget that Argentina is so scarcely populated that we don't have a railway to the South. And Tierra del Fuego is an island, so you'd need more tunnels (keep the zeros goings!).
BTW, if the cost of it were like U$S 250M I'm sure our gov would quote U$S 2000M (8:1 rule of corruption). But I've heard that Russian is still more "inefficient" (I would guess it's 40:1, but I don't have enough experience with that gov).
May you point out _which_ were those failed attempts? Cause I'm being a bit skeptical about your frasing.
Well, I wouldn't call the i740 bleeding edge, the i810 a price/performance champion, the i820 or i840 a robust platform. Recall the how late the P!!! and 820 were (reliability does includes promised launch dates). Consider how the crippled the Celeron by keeping its FSB at 66Mhz. Recall how they had to cede the whole server/workstation chipset market to ServerWorks just because they had to use RDRAM. No, I'm not the typical slashdotter and yet I've lost all my respecto for Intel.