At a physical level, peering connections look exactly like any other type of connection. The diffrence is in how it is configured. With peering, you only get the routes of the network you are connected to. So, if I am peered with Sprint, I only get Sprint routes, and Sprint's Customers routes. Thats all, nothing else.
A little peering history. Your government mandated public peering points into existence to try and free up the market, to help get a little competition cooking. however, technology had a hard time keeping up with demand at these internet hot spots, and public peering points soon became a serious traffic loss area for the bigger networks. (After all, with every moe, dick and larry trying to funnel traffic down that DS3 or FDDI connection at the NAP, it soon filled up!) So anyone who seriously needed bandwidth to larger networks ended up privatly peering (your 2 shacks in the woods). Of course, private peering is a lot more expensive for both parties, so the larger networks eventualy relized that no-cost peering with the smaller networks was of little or no value to them. After all, if you are UUnet peering with Joe Small Rocks ISP, he gets free access to your 100k worth of customers, and you get access to his 10 customers!
It basicly boiled down to giving away free service to smaller ISPs for no value in return, so the larger ISPs started coming up with a set of requirments to keep the smaller ISPs from wasting thier resources.
Things like : Be at 5 public access peering points, have a presence in XX number of states, document XXXXMb of traffic between our networks for XXhours of uptime. In other words, make us understand how it is to our mutual advantage to peer. (after all, peering relationships shoule be among peers, not big network to little network)
You are pretty much right, the almighty dollar rules here. But, that is as it should be. We don't want the Internet to become run on the same rules as a welfare state. I believe such rules would result in the least stable service for a relativly high amount of money.
So, bottom line - if you want no-cost peering, have a big network with lots of customers! Plan on burning a lot of money up front to build a big network and even more money to BUY transit until you have enough customers you can meet the demands for private peering. Once you meet those demands, you will get it.
I hope someone prints this out and sells this entire thread to one of those 'growing up catholic' books. If there is one things Catholics are good at, it's laughing at themselves, and how other people choose to precieve them!
This thread has made me laugh harder than anything in recent memory.
Heheh. The Church has more practice at ALL of the above tactics than most governments. It HAS done all of the things you have mentioned in the past.
The diffrence now is that the Pope frowns on such a misuse of power.
I can see it now: Danni's Hard Drive 1st site to be digitaly-excommunicated by the Roman Catholic Church.
I think the thing that might scare you is the concept of someone not being driven by a profit motive. People who are after money are easy to predict and control.
This is a good point, and it points out one of the fundamental problems with any kind of filtering technology: You can't replace mommy and daddy with a piece of software.
however, I would also say the Church has every right to filter content on thier network in any way they see fit. You have to remember, this is a two pronged service: 1) Provide access to those who can't afford to pay for it and 2) Promote a christian/catholic view of the universe.
As always, if you become a 'victim' of a filter, the easist way to deal with it is to TALK to the admins! As long as you are dealing with a contientious bunch (hmm, could you call the church that?) they should be more than happy to fix any problems you run into.
Cogent is useing WCG fiber for thier backbone, another company in the same $1000 per 100Meg connection business is Yipes!. The basis for this business plan is quite simple: Crossconnects for T based services are expensive and difficult to maintain. Ethernet is cheap and easy to maintain. By elimintating the whole 'telco style' cross connect system, you can save a HUGE amount of money. Yipes! seems to be smart enough to relize they need to be able to charge on a per meg notch, and are using 'smart' ethernet switches (I think Extreme) to handle provisioning and bandwidth control. Cogent is doing something similar.
Both these companies have a sound plan, they have rubbed thier crystal balls and seen into the future. They have both made the mental leap of faith required. Now it's just a matter of who can grow it the fastest.
I know this has been aroud for an increadibly long time...
ftp://research.att.com/dist/mab/cfs.ps
and I think they have also built a kernel module to do it at the kernel level (cfs looks like an NFS file mount, to get around all that nast kernal stuff), but I don't have the time to go hunting for it...
Yet another 'feature' that is Brand New(tm) now that Slashdot has 'discovered' it.
Being a technical manager can be a lot of fun, if you think like a General (or even a Captain or a Colonel). Even though you have to accept some additional responsobility, the paperwork doesn't have to be a killer, just make sure they let you hire a good admin assistent to take care of that stuff. That will free you up to actually ENJOY steering a group of really smart guys into acomplishing more than the sum of thier parts.
I can NOT stress enough, that if you remain unorganized, and don't PLAN for things to happen, you will be a crappy manager, and no one will want to work for you. A good manager's FIRST rule (at least if you manage talented technical types)
is that YOU work for THEM, not the other way around. This give you the chance to lead them in the direction/towards acomplishing the goal the YOUR boss wants done. Not to drop into sales droid speak, but that would be a WIN/WIN situation, the best anyone can hope for.
Just because a lot of the more technicaly gifted people that post here don't have a taste for the responsibility and challange of being in charge, don't let them frighten you off (unless you really believe you CAN'T do it.) There is NOTHING wrong with admiting you can't do something before you royaly screw the pooch.
If the recording industry REALLY wants to put a dent in the way people share files via Napster, they ought to try :
1) Putting there own, high speed server in the Napster system with 10-30sec of music, followed by an 'ad' on where to get the 'real thing'
2) Try and convince napster to put a file size limit on shared files.
Granted, 2 would probably end up making people go to Gnutella or something (which would solve Napster's problems as well, but not in a way I think they would like)
And he is probably wrong. But because the people who are going to be building the big networks come from a 'phone company' mentality, that is probably what we are going to end up with.
Could you imagine a 'profit sharing' model between the network provider and the content provider?
That wasn't possible with voice and LD, but it is with the Internet!
Ass-umming the PR/Marketing people are allowed to make the priceing schemes, I think you will see more of a 'we give you the access or means to get access, but you have to use our services' model, similar to what the phone company used to do with Voice services. (ie, they would LEASE you the telephone, and you would pay them for the hardware as well as the service)
You dont need QoS to do that, TCP does it automaticly.
QoS is what you make of it. Although for the moment, your options on most IP hardware consist of:
Pass it
Please do a little more research into what the 'State-of-The-Art' technology for IP traffic is before you go trawling for opinions.
There are only TWO things that keep you from haveing 100BaseT to your house to the backbone: 1) Money and 2)Corporate Vision.
The technology is certainly there, and has been for some time. Do a little traffic study on how the traffic flows, and you would notice that Akamai, I-Beam and others are doing wonders to help large backbones NOT have to transit a ton of traffic.
Look at some of the newer access control systems like Nortel's Shasta 5000 (32k firewalls, all at broadband speed!) Or Lucent's Springtide box. These have been SHIPPING all this year (and in some cases even the middle of last year)
Look at what is SHIPPING now for Core IP switching technologies: Every 'next-gen' terrabit switch/router is offering 16 X OC-192 (Thats 160 GIGS) speeds accross multiple optical connections (obviously designed for use with DWDM to cut fiber costs)
Next Gen Optical gear is pushing toward several dozen OC-192s per STRAND!
Williams,Enron,Level 3, and Qwest are ALL laying 96 strand fiber trunks (in some cases 3 or 4 of them) in the ground as fast as the trenchers can get out there.
Do the math. All it takes is for these companies to have a little bit of vision.
of Course, I notice we missed the HUGE downside to more bandwidth. Something I have been moaning about for the last 2 year I call 'Protocol Bloat' If you think it sounds like code bloat, you are RIGHT, just as more Drive Space and Memory have become available (causing code to loose efficiency), so too will haveing a glut of bandwidth cause the applications and protocols used on that system to become foolishly (and perhaps wrecklessly) unwieldly.
I can see 5 and 6 Meg 'Jave Applets' just to view a certain sites content... IP 'tunneled' to the point of insanity for the sake of 'copy protection'
This is the best truth I have heard in this discussion so far. SO many 'guidance councilors' these days send the confused and greedy in a direction not truly suited to them. In the next few years, I am afraid the quality of IT professional we will be seeing turned out of these grand old institutions will be dropping, if the prime motivator is money. Claudius, you have the truth of it there, school is for scholarship, not trade preparation. In truth most of the scholarly people I know arn't fit to run a network, but they get great joy from learning and teaching! They understand the WHAT and the WHY of things, but arn't really all that interested in the HOW. (Thats what technicians and sysadmins are for)
If your just in school to increase your chances at getting a high paying job, do us all a favor and evaluate if you are going to:
1) Actually enjoy what it is you plan on doing when you get your degree.
2) Are actually any GOOD at what you plan on doing when you get your degree.
The cream floats to the top, no matter what. Degree or no Degree, if you ENJOY doing something and are GOOD at it, you WILL find a way to get paid for it.
and top dollar too, (BTW, if you are GOOD, then a smart company will see it's way clear to overlooking just about anything, after all, it's about money. The aura of privledge companies that will NOT look at your acomplishments and skills vs the completion of a degree, would NOT sit well with someone who is a 'self made man/woman'. In other words, if they won't take you, you didn't want to work there anyway.:> )
-Not enough good Mirrors : What network are you hooked up through? I can honestly say, anytime I have problems getting to a mirror, it is because of my network provider, not the quality or quantity of the mirror sites.
-Fetch is broken (broken/truncated files) - See 1st question.
- Compile from sources only -/stand/sysinstall has a binary package install utility built in.
[Shrug] I don't see what your beef is.
Ack, replying to my own post, It DOES look like I2 is doing 'internet speed record' testing, seems like they topped out at about 8Gig... So, maybe they ARE getting the 192 goodies... Of course, now you have to build a test set that can generate that much traffic:>
When you have the larger global/national carriers putting in 10Gig links (OC192) to cope with the current demands, and switch vendors building switches that can handle multiple 10Gig links as a single path (read Multi-Link PPP writ LARGE), running a backbone with some tiny little OC48s (2.5 Gig) doesn't seem all that impressive. Granted, it was the technology tester that helped us get to where we are now, but notice that I2 isn't getting the 192 links, The 'Real' Internet is.
Anybody remember that old TV show with Richard Dean Anderson? Wouldn't it be cool to have a game where the hero gets out of sticky situations without having to resort to the full auto Uzi?
Slashdot Mirror
At a physical level, peering connections look exactly like any other type of connection. The diffrence is in how it is configured. With peering, you only get the routes of the network you are connected to. So, if I am peered with Sprint, I only get Sprint routes, and Sprint's Customers routes. Thats all, nothing else.
A little peering history. Your government mandated public peering points into existence to try and free up the market, to help get a little competition cooking. however, technology had a hard time keeping up with demand at these internet hot spots, and public peering points soon became a serious traffic loss area for the bigger networks. (After all, with every moe, dick and larry trying to funnel traffic down that DS3 or FDDI connection at the NAP, it soon filled up!) So anyone who seriously needed bandwidth to larger networks ended up privatly peering (your 2 shacks in the woods). Of course, private peering is a lot more expensive for both parties, so the larger networks eventualy relized that no-cost peering with the smaller networks was of little or no value to them. After all, if you are UUnet peering with Joe Small Rocks ISP, he gets free access to your 100k worth of customers, and you get access to his 10 customers!
It basicly boiled down to giving away free service to smaller ISPs for no value in return, so the larger ISPs started coming up with a set of requirments to keep the smaller ISPs from wasting thier resources.
Things like : Be at 5 public access peering points, have a presence in XX number of states, document XXXXMb of traffic between our networks for XXhours of uptime. In other words, make us understand how it is to our mutual advantage to peer. (after all, peering relationships shoule be among peers, not big network to little network)
You are pretty much right, the almighty dollar rules here. But, that is as it should be. We don't want the Internet to become run on the same rules as a welfare state. I believe such rules would result in the least stable service for a relativly high amount of money.
So, bottom line - if you want no-cost peering, have a big network with lots of customers! Plan on burning a lot of money up front to build a big network and even more money to BUY transit until you have enough customers you can meet the demands for private peering. Once you meet those demands, you will get it.
Heh, Zephyr, the grand daddy of IM. Don't you need Kerberos for zephyr to work? Or has someone updated it?
Sure they are. They just don't spend a lot of time advertising it.
I hope someone prints this out and sells this entire thread to one of those 'growing up catholic' books. If there is one things Catholics are good at, it's laughing at themselves, and how other people choose to precieve them!
This thread has made me laugh harder than anything in recent memory.
Heheh. The Church has more practice at ALL of the above tactics than most governments. It HAS done all of the things you have mentioned in the past.
The diffrence now is that the Pope frowns on such a misuse of power.
I can see it now: Danni's Hard Drive 1st site to be digitaly-excommunicated by the Roman Catholic Church.
I think the thing that might scare you is the concept of someone not being driven by a profit motive. People who are after money are easy to predict and control.
This is a good point, and it points out one of the fundamental problems with any kind of filtering technology: You can't replace mommy and daddy with a piece of software.
however, I would also say the Church has every right to filter content on thier network in any way they see fit. You have to remember, this is a two pronged service: 1) Provide access to those who can't afford to pay for it and 2) Promote a christian/catholic view of the universe.
As always, if you become a 'victim' of a filter, the easist way to deal with it is to TALK to the admins! As long as you are dealing with a contientious bunch (hmm, could you call the church that?) they should be more than happy to fix any problems you run into.
Use Yipes! in houston. They have a very similar deal.
They don't need to peer, they only need to buy transit. Peering comes a few months after they fill up the transit lines :>
Cogent is useing WCG fiber for thier backbone, another company in the same $1000 per 100Meg connection business is Yipes!. The basis for this business plan is quite simple: Crossconnects for T based services are expensive and difficult to maintain. Ethernet is cheap and easy to maintain. By elimintating the whole 'telco style' cross connect system, you can save a HUGE amount of money. Yipes! seems to be smart enough to relize they need to be able to charge on a per meg notch, and are using 'smart' ethernet switches (I think Extreme) to handle provisioning and bandwidth control. Cogent is doing something similar. Both these companies have a sound plan, they have rubbed thier crystal balls and seen into the future. They have both made the mental leap of faith required. Now it's just a matter of who can grow it the fastest.
Sorry the blowfish encrypted swap filesystem is in OPENBSD...
Also check out TCFS http://tcfs.dia.unisa.it/group-sharing.html
I know this has been aroud for an increadibly long time...
ftp://research.att.com/dist/mab/cfs.ps
and I think they have also built a kernel module to do it at the kernel level (cfs looks like an NFS file mount, to get around all that nast kernal stuff), but I don't have the time to go hunting for it...
Yet another 'feature' that is Brand New(tm) now that Slashdot has 'discovered' it.
You can use ODIGO . It speaks ICQ and AIM, runs on windows though. (Unix/linux has Everybuddy, what an easy thing to install on my BSD box :> )
Don't be fooled!
Being a technical manager can be a lot of fun, if you think like a General (or even a Captain or a Colonel). Even though you have to accept some additional responsobility, the paperwork doesn't have to be a killer, just make sure they let you hire a good admin assistent to take care of that stuff. That will free you up to actually ENJOY steering a group of really smart guys into acomplishing more than the sum of thier parts.
I can NOT stress enough, that if you remain unorganized, and don't PLAN for things to happen, you will be a crappy manager, and no one will want to work for you. A good manager's FIRST rule (at least if you manage talented technical types)
is that YOU work for THEM, not the other way around. This give you the chance to lead them in the direction/towards acomplishing the goal the YOUR boss wants done. Not to drop into sales droid speak, but that would be a WIN/WIN situation, the best anyone can hope for.
Just because a lot of the more technicaly gifted people that post here don't have a taste for the responsibility and challange of being in charge, don't let them frighten you off (unless you really believe you CAN'T do it.) There is NOTHING wrong with admiting you can't do something before you royaly screw the pooch.
Good Luck!
If the recording industry REALLY wants to put a dent in the way people share files via Napster, they ought to try :
1) Putting there own, high speed server in the Napster system with 10-30sec of music, followed by an 'ad' on where to get the 'real thing'
2) Try and convince napster to put a file size limit on shared files.
Granted, 2 would probably end up making people go to Gnutella or something (which would solve Napster's problems as well, but not in a way I think they would like)
And he is probably wrong. But because the people who are going to be building the big networks come from a 'phone company' mentality, that is probably what we are going to end up with.
Could you imagine a 'profit sharing' model between the network provider and the content provider?
That wasn't possible with voice and LD, but it is with the Internet!
Ass-umming the PR/Marketing people are allowed to make the priceing schemes, I think you will see more of a 'we give you the access or means to get access, but you have to use our services' model, similar to what the phone company used to do with Voice services. (ie, they would LEASE you the telephone, and you would pay them for the hardware as well as the service)
You dont need QoS to do that, TCP does it automaticly.
QoS is what you make of it. Although for the moment, your options on most IP hardware consist of:
Pass it
Drop it
Queue it
Check out Yipes!, they are estimating prices of about $1200 a month for a 100BaseT connection...
(of course, only in cities where they have a POP)
Please do a little more research into what the 'State-of-The-Art' technology for IP traffic is before you go trawling for opinions.
There are only TWO things that keep you from haveing 100BaseT to your house to the backbone: 1) Money and 2)Corporate Vision.
The technology is certainly there, and has been for some time. Do a little traffic study on how the traffic flows, and you would notice that Akamai, I-Beam and others are doing wonders to help large backbones NOT have to transit a ton of traffic.
Look at some of the newer access control systems like Nortel's Shasta 5000 (32k firewalls, all at broadband speed!) Or Lucent's Springtide box. These have been SHIPPING all this year (and in some cases even the middle of last year)
Look at what is SHIPPING now for Core IP switching technologies: Every 'next-gen' terrabit switch/router is offering 16 X OC-192 (Thats 160 GIGS) speeds accross multiple optical connections (obviously designed for use with DWDM to cut fiber costs)
Next Gen Optical gear is pushing toward several dozen OC-192s per STRAND!
Williams,Enron,Level 3, and Qwest are ALL laying 96 strand fiber trunks (in some cases 3 or 4 of them) in the ground as fast as the trenchers can get out there.
Do the math. All it takes is for these companies to have a little bit of vision.
of Course, I notice we missed the HUGE downside to more bandwidth. Something I have been moaning about for the last 2 year I call 'Protocol Bloat' If you think it sounds like code bloat, you are RIGHT, just as more Drive Space and Memory have become available (causing code to loose efficiency), so too will haveing a glut of bandwidth cause the applications and protocols used on that system to become foolishly (and perhaps wrecklessly) unwieldly.
I can see 5 and 6 Meg 'Jave Applets' just to view a certain sites content... IP 'tunneled' to the point of insanity for the sake of 'copy protection'
This is the best truth I have heard in this discussion so far. SO many 'guidance councilors' these days send the confused and greedy in a direction not truly suited to them. In the next few years, I am afraid the quality of IT professional we will be seeing turned out of these grand old institutions will be dropping, if the prime motivator is money.
:> )
Claudius, you have the truth of it there, school is for scholarship, not trade preparation. In truth most of the scholarly people I know arn't fit to run a network, but they get great joy from learning and teaching! They understand the WHAT and the WHY of things, but arn't really all that interested in the HOW. (Thats what technicians and sysadmins are for)
If your just in school to increase your chances at getting a high paying job, do us all a favor and evaluate if you are going to:
1) Actually enjoy what it is you plan on doing when you get your degree.
2) Are actually any GOOD at what you plan on doing when you get your degree.
The cream floats to the top, no matter what. Degree or no Degree, if you ENJOY doing something and are GOOD at it, you WILL find a way to get paid for it.
and top dollar too, (BTW, if you are GOOD, then a smart company will see it's way clear to overlooking just about anything, after all, it's about money. The aura of privledge companies that will NOT look at your acomplishments and skills vs the completion of a degree, would NOT sit well with someone who is a 'self made man/woman'. In other words, if they won't take you, you didn't want to work there anyway.
-Not enough good Mirrors : What network are you hooked up through? I can honestly say, anytime I have problems getting to a mirror, it is because of my network provider, not the quality or quantity of the mirror sites. /stand/sysinstall has a binary package install utility built in.
-Fetch is broken (broken/truncated files) - See 1st question.
- Compile from sources only -
[Shrug] I don't see what your beef is.
Ack, replying to my own post, It DOES look like I2 is doing 'internet speed record' testing, seems like they topped out at about 8Gig... So, maybe they ARE getting the 192 goodies... Of course, now you have to build a test set that can generate that much traffic :>
When you have the larger global/national carriers putting in 10Gig links (OC192) to cope with the current demands, and switch vendors building switches that can handle multiple 10Gig links as a single path (read Multi-Link PPP writ LARGE), running a backbone with some tiny little OC48s (2.5 Gig) doesn't seem all that impressive. Granted, it was the technology tester that helped us get to where we are now, but notice that I2 isn't getting the 192 links, The 'Real' Internet is.
Anybody remember that old TV show with Richard Dean Anderson? Wouldn't it be cool to have a game where the hero gets out of sticky situations without having to resort to the full auto Uzi?
Nah, it would never sell
Who said it had to be a cohesive story? Ever try a good old fashioned MUD?