You'd be horrified to know how many "dumb" customers what to run BGP when they have no need for it and certainly no clue how. (And most of those idiots don't have a router that could handle a BGP process much less a full route table. [my table is 24-28M at the moment depending on the router]) They also fail to understand what it will take to get an ASN -- you think IP addresses are scarce; BGP ASNs are very limited.
That's why the radb exists. (it came into being to serve as a central knowledge base of who controls what to simplify routing management. one could update the radb and all the ISPs building rules based on it would be updated automagically without 37 individual emails to the wrong people.)
And how is this any different from current filtering practices? (as-path and distribute-list filters which even the dumbest ISPs have been doing for years.) All the signing and "authorisation" does is increase processing loads and memory utilization, and require some group of central authorities to be the root of the trust tree. For this sort of thing to work, every prefix would have to be signed with a unique key.
This isn't something that can be fixed on napkins over dinner. And it's technically not something that needs new (read: overly complex) technology to fix. If the '97 incident was as serious as the article indicates, why don't I remember it? I remember some idiots stealing 199.72.1.0/24 and pissing us off for a few days. BGP has long since ignored 0/0 (unless instructed otherwise), so "[advertising] best route to the entire internet" would be over 120,000 route entries at this very moment. (things to be considered... as-path length, prefix-length, metrics which nobody sends, and locally configured preferences.)
The IBM is an xSeries server under contract. They will replace the entire server; not any component within it. Yes, that's fucking stupid. The other engineers won't let me near it with a screw driver.
You obviously haven't taken a modern Sun apart. The MBs are built with almost identical specs. The cases are made of the same metal and plastic of the same strengths. No, Sun doesn't assemble systems in 5$ zinc cases, but no other major vender does either. There's no point in armor plating a server that's going to spend it's life bolted into a rack -- it's not like it's being mounted to the roof of tank.
Don't get me wrong, Sun makes good stuff. It's just not worth the cost. I've gotten just as high a reliablity from x86 hardware at a fraction of the cost.
... and additionally, that's not "IN a PC". Any idiot can bolt a bunch of drives to a rail and sit it on a table. (My former boss did and then proceeded to whine about how the 25$ hpt controllers didn't work correctly -- BIOS limit was 3 in a system, w2k driver was limited to 1. 3ware to the rescue!)
And exactly how was less not functional? It'll work without any terminal capabilities:
# uname -a SunOS XXX 5.8 Generic_108528-19 sun4u sparc SUNW,Ultra-80 # which less/usr/bin/less # less -V less 340 Copyright (C) 1999 Mark Nudelman
less comes with NO WARRANTY, to the extent permitted by law. You may redistribute copies of less under the terms of the GNU General Public License. For more information about these matters, see the file named COPYING in the less distribution. # unsetenv TERM # less/etc/path_to_inst WARNING: terminal is not fully functional #etc/path_to_inst (press RETURN) # Caution! This file contains critical kernel state #...
You missed the point. There's nothing about a Sun Fire that makes it more stable or reliable than PC server hardware -- component failures result in crashes and happen just as often (read: rarely) in both camps. (And I speak from decades of experience.) That being said, there's nothing about the servers being assembled by Dell, Gateway, and others that you cannot build yourself for a fraction of the cost.
For year, I've been appauled by the money companies waste on name brands and support contracts. (And that's not including the hoops they jump through in executing their support options. For example, IBM won't send us a replacement for a failed hard drive; we are required to ship the entire server back which can not happen.)
Umm, I've sold similar systems with large storage arrays for 1/3 that cost. [dual 2.4G Xeon, 2G RAM, 600G RAID5, database/web server with their custom applications integrated, hand delivered and installed to Deposit, NY -- during a snow storm no less.]
And given some random computing task, on average, the Xeon system will very likely run circles around the SunFire.
Sun licensed the trademark -- I don't think AT&T sold it to them out-right. So therefore, Solaris is UNIX (tm).
And if you insist on AT&T as the only UNIX (tm), then the last time I used UNIX (tm) was a few days ago... AT&T SVR4 is the central OS for the Lucent 5ESS phone switch.
Now, let's be fair here. He changed both the OS AND the platform -- solaris/sparc -> linux/x86. In this case, the choice of OS didn't make much of a difference. x86 hardware is much cheaper and insanely faster.
There are advantages to sparc hardware, but not many; and they are certainly not cost effective.
If that's 3TB of SCSI storage, then it might be note worthy. But it's certainly not a 6 o'clock news event.
Why is this news anyway? I, personally, have built (and sold) several 1TB+ "PCs" over the last few years. 1.8TB can be done with a half dozen drives these days. (for the cost of *2* large SCSI drives, even.) Heh, I could fit that in a 25$ mini-tower case.
you will often get "Jargon Monkeys" who can b.s. their way into the position
Not with me interviewing them. When we were looking for a "networking monkey", at the end of the BS part, we'd set the interviewee infront of a few routers and see what he'd do with them... and no, they weren't aware ahead of time that we were going to have them setup some routers. Only one of the people didn't have a mini-nervous breakdown at the idea. (He went straight to OSPF... He's also the only one to ask how we had IOS 12.0 running on a Cisco AccessPro -- the 2500 in a white case, the oldest thing we could find.)
Well, he's also missed the mark on the salaries he's listing. 75k for a windows admin is rather high and actually insanely high for a university -- try 50-60k in the current markets (and it depends on the location.) 120k for a UNIX admin is just absurd -- esp. when teenagers can do it for minimum wage. I guess he's unaware of the SANS yearly salary survey.
Indeed. Has the American gene pool become severely compromised and poluted over the last decade or two? Seeing how one generation is between 20 and 30 years, that's very unlikely. The levels of violence in society has increased orders of magnitude over a very short period of time -- much to short a period to be attributable to genetics. The only thing I can point to is a serious decline in "parenthood"; parents cannot be bothered to raise their children...
20 to 30 years ago, we weren't a bunch of gun-toting, troll-like, social rejects snapping like dry twigs and blowing up malls. Back then, children were punished for misbehaving -- and I'm not talking about the lame, ineffectual "timeout". Parents actually took time to know what their children were doing. And they knew where we were -- if we weren't where we were supposed to be, out came the punishment.
In the modern world, it's easier to make up some "disorder" to explain people's behavior than address the behavior. Part of it is the greed of doctors, clinics, and entire industries. But the other part is the parents clinging to an "it's not my fault" line -- there's no way their lack of parenting is to blame for their child's ADD/OCD/etc. It's bullshit. Bullshit I tell you! Most people sucked their thumb as a child. That's technically an OCD, but that habit is broken after a few years. (either by "old school" punishment or modern passive "shame" methods.)
That depends on one's classification of significant. I, and in fact, most of world, will differ with you on this.
If you really think there's a bug you should report [it]
99% of the time, that's useless. I'm convinced it takes a nuclear armed army to even get MS to admit something is borked. They will repeatedly claim the user doesn't know what they are doing or "it's not a bug." Even after being given the precise steps to reproduce the bug, they will say it's not broken. When pushed, they simply respond with "you don't have a support contract" and essentially "fuck off."
[Case in point: I tried repeatedly to report a caching bug in all versions of IE... try it yourself. IE will not cache anything to the host "www.hp.com" With a simple trailing dot ("www.hp.com.") IE goes to the exact same place and fetches the exact same data and caches it.]
We don't do a new version to fix bugs.
Ok. New versions are not released solely to fix bugs. However, in many cases, the only way to get certain bugs fixed is to buy the new version. Microsoft is alone in this arena.
Guess how much we spend on phone calls every year
An insignificant fraction of the revenue generated by those calls -- add up all the support contracts and 250$/incident costs... And, btw, people aren't going to call to report a bug; they send lengthly, detailed emails with attachments. Explaining a bug to the sheep answering the support lines is a lengthly waste of time.
We've never done a piece of software unless we thought it would sell
Right. So what moron thought Bob would sell?
And let's be more accurate, Microsoft doesn't do anything unless someone else is making money doing it. (double space, TCP/IP (networking in general), disk defrag tools, terminal server, the entire office product suite...)
For actual board layout, yes, that's scripted. But the actual process of designing a circuit is not done with vi; it's done in the schematic designer GUI.
I guess you'd call me "very old school"... I prefer pencil and paper to any software tools. I can do things much faster and with a great deal less stress by hand. (with the exception of PCB layout. that takes forever by hand.)
Blocking the actual 911 call... I seriously doubt it. That's programmed into a phone switch that isn't connected to the internet (very, very few switches can be directly connected to the net.) Disruption to the E911 call center is a very likely. Someone will answer; they just won't be able to do anything.
The 10/8 addresses are perfectly ok within the RR network. However, they should never be allowed to leave that network. In order for me to see those ICMP replies (TTL exceed, btw), they have to leave RR, pass through ATDN, to Sprint, to my provider, to me.
Pay attention to this part: A packet with a source address in a private (non-routed) IP range, passed through three (3) ISPs (2 of them teir1 providers) which means none of the them are lifting a finger to prevent spoofing.
It's not so much a matter of "knowing"... A lot of the stupid stuff you'll find (and if you actually go looking for it...) that is simply the result of "lazy" contractors.
As the saying goes, "Close enough for government work."
And just how does that differ from the way InterNIC used to manage domains? (Answer: it doesn't)
Obviously, no one read the part about an email address within NIPRnet. If you have an email address within NIPRnet, then you work for the DoD. It's not like someone can use foo@bar.com to register dozens of domains.
And an other thing, all the web engine does is fillout the templates that have been available via ftp for years.
(I'm assuming Google has already removed the reffered to pages from the cache.)
... because we all know the private address ranges are only unsed by end users behind NAT and firewalls.
Have you ran a traceroute to a Road Runner cable modem lately? 15 66.26.33.179 33.238 ms 32.589 ms 32.510 ms 16 24.25.1.101 34.965 ms 33.925 ms 34.576 ms 17 10.41.96.220 40.359 ms 40.125 ms 39.868 ms 18...
(Hop 17 is a Cisco uBR925 owned by RR.)
The same was true for CAIS DSL service too -- they used the 172.16... range.
Slashdot Mirror
You'd be horrified to know how many "dumb" customers what to run BGP when they have no need for it and certainly no clue how. (And most of those idiots don't have a router that could handle a BGP process much less a full route table. [my table is 24-28M at the moment depending on the router]) They also fail to understand what it will take to get an ASN -- you think IP addresses are scarce; BGP ASNs are very limited.
See also: as-path
:-) [I've been there. It isn't that easy.]
It's already trackable. The real problem is getting the nearest peer to "turn the idiots off".
s/RIPE/radb/
That's why the radb exists. (it came into being to serve as a central knowledge base of who controls what to simplify routing management. one could update the radb and all the ISPs building rules based on it would be updated automagically without 37 individual emails to the wrong people.)
And how is this any different from current filtering practices? (as-path and distribute-list filters which even the dumbest ISPs have been doing for years.) All the signing and "authorisation" does is increase processing loads and memory utilization, and require some group of central authorities to be the root of the trust tree. For this sort of thing to work, every prefix would have to be signed with a unique key.
This isn't something that can be fixed on napkins over dinner. And it's technically not something that needs new (read: overly complex) technology to fix. If the '97 incident was as serious as the article indicates, why don't I remember it? I remember some idiots stealing 199.72.1.0/24 and pissing us off for a few days. BGP has long since ignored 0/0 (unless instructed otherwise), so "[advertising] best route to the entire internet" would be over 120,000 route entries at this very moment. (things to be considered... as-path length, prefix-length, metrics which nobody sends, and locally configured preferences.)
The IBM is an xSeries server under contract. They will replace the entire server; not any component within it. Yes, that's fucking stupid. The other engineers won't let me near it with a screw driver.
You obviously haven't taken a modern Sun apart. The MBs are built with almost identical specs. The cases are made of the same metal and plastic of the same strengths. No, Sun doesn't assemble systems in 5$ zinc cases, but no other major vender does either. There's no point in armor plating a server that's going to spend it's life bolted into a rack -- it's not like it's being mounted to the roof of tank.
Don't get me wrong, Sun makes good stuff. It's just not worth the cost. I've gotten just as high a reliablity from x86 hardware at a fraction of the cost.
... and additionally, that's not "IN a PC". Any idiot can bolt a bunch of drives to a rail and sit it on a table. (My former boss did and then proceeded to whine about how the 25$ hpt controllers didn't work correctly -- BIOS limit was 3 in a system, w2k driver was limited to 1. 3ware to the rescue!)
SunOS XXX 5.8 Generic_108528-19 sun4u sparc SUNW,Ultra-80
# which less
# less -V
less 340
Copyright (C) 1999 Mark Nudelman
less comes with NO WARRANTY, to the extent permitted by law.
You may redistribute copies of less under the terms
of the GNU General Public License.
For more information about these matters,
see the file named COPYING in the less distribution.
# unsetenv TERM
# less
WARNING: terminal is not fully functional
#etc/path_to_inst (press RETURN)
# Caution! This file contains critical kernel state
#
You missed the point. There's nothing about a Sun Fire that makes it more stable or reliable than PC server hardware -- component failures result in crashes and happen just as often (read: rarely) in both camps. (And I speak from decades of experience.) That being said, there's nothing about the servers being assembled by Dell, Gateway, and others that you cannot build yourself for a fraction of the cost.
For year, I've been appauled by the money companies waste on name brands and support contracts. (And that's not including the hoops they jump through in executing their support options. For example, IBM won't send us a replacement for a failed hard drive; we are required to ship the entire server back which can not happen.)
Umm, I've sold similar systems with large storage arrays for 1/3 that cost. [dual 2.4G Xeon, 2G RAM, 600G RAID5, database/web server with their custom applications integrated, hand delivered and installed to Deposit, NY -- during a snow storm no less.]
And given some random computing task, on average, the Xeon system will very likely run circles around the SunFire.
Sun licensed the trademark -- I don't think AT&T sold it to them out-right. So therefore, Solaris is UNIX (tm).
And if you insist on AT&T as the only UNIX (tm), then the last time I used UNIX (tm) was a few days ago... AT&T SVR4 is the central OS for the Lucent 5ESS phone switch.
Now, let's be fair here. He changed both the OS AND the platform -- solaris/sparc -> linux/x86. In this case, the choice of OS didn't make much of a difference. x86 hardware is much cheaper and insanely faster.
There are advantages to sparc hardware, but not many; and they are certainly not cost effective.
This has everything to do with more. On many systems, more will not page backwards from stdin.
Adjust your environment to include a "setenv PAGER less" (or equiv.) and be done with it. Or replace more with a link to less.
If that's 3TB of SCSI storage, then it might be note worthy. But it's certainly not a 6 o'clock news event.
Why is this news anyway? I, personally, have built (and sold) several 1TB+ "PCs" over the last few years. 1.8TB can be done with a half dozen drives these days. (for the cost of *2* large SCSI drives, even.) Heh, I could fit that in a 25$ mini-tower case.
- you will often get "Jargon Monkeys" who can b.s. their way into the position
Not with me interviewing them. When we were looking for a "networking monkey", at the end of the BS part, we'd set the interviewee infront of a few routers and see what he'd do with them... and no, they weren't aware ahead of time that we were going to have them setup some routers. Only one of the people didn't have a mini-nervous breakdown at the idea. (He went straight to OSPF... He's also the only one to ask how we had IOS 12.0 running on a Cisco AccessPro -- the 2500 in a white case, the oldest thing we could find.)Well, he's also missed the mark on the salaries he's listing. 75k for a windows admin is rather high and actually insanely high for a university -- try 50-60k in the current markets (and it depends on the location.) 120k for a UNIX admin is just absurd -- esp. when teenagers can do it for minimum wage. I guess he's unaware of the SANS yearly salary survey.
Indeed. Has the American gene pool become severely compromised and poluted over the last decade or two? Seeing how one generation is between 20 and 30 years, that's very unlikely. The levels of violence in society has increased orders of magnitude over a very short period of time -- much to short a period to be attributable to genetics. The only thing I can point to is a serious decline in "parenthood"; parents cannot be bothered to raise their children...
20 to 30 years ago, we weren't a bunch of gun-toting, troll-like, social rejects snapping like dry twigs and blowing up malls. Back then, children were punished for misbehaving -- and I'm not talking about the lame, ineffectual "timeout". Parents actually took time to know what their children were doing. And they knew where we were -- if we weren't where we were supposed to be, out came the punishment.
In the modern world, it's easier to make up some "disorder" to explain people's behavior than address the behavior. Part of it is the greed of doctors, clinics, and entire industries. But the other part is the parents clinging to an "it's not my fault" line -- there's no way their lack of parenting is to blame for their child's ADD/OCD/etc. It's bullshit. Bullshit I tell you! Most people sucked their thumb as a child. That's technically an OCD, but that habit is broken after a few years. (either by "old school" punishment or modern passive "shame" methods.)
- There are no significant bugs
...
That depends on one's classification of significant. I, and in fact, most of world, will differ with you on this.- If you really think there's a bug you should report [it]
99% of the time, that's useless. I'm convinced it takes a nuclear armed army to even get MS to admit something is borked. They will repeatedly claim the user doesn't know what they are doing or "it's not a bug." Even after being given the precise steps to reproduce the bug, they will say it's not broken. When pushed, they simply respond with "you don't have a support contract" and essentially "fuck off."[Case in point: I tried repeatedly to report a caching bug in all versions of IE... try it yourself. IE will not cache anything to the host "www.hp.com" With a simple trailing dot ("www.hp.com.") IE goes to the exact same place and fetches the exact same data and caches it.]
- We don't do a new version to fix bugs.
Ok. New versions are not released solely to fix bugs. However, in many cases, the only way to get certain bugs fixed is to buy the new version. Microsoft is alone in this arena.- Guess how much we spend on phone calls every year
An insignificant fraction of the revenue generated by those calls -- add up all the support contracts and 250$/incident costs... And, btw, people aren't going to call to report a bug; they send lengthly, detailed emails with attachments. Explaining a bug to the sheep answering the support lines is a lengthly waste of time.- We've never done a piece of software unless we thought it would sell
Right. So what moron thought Bob would sell?And let's be more accurate, Microsoft doesn't do anything unless someone else is making money doing it. (double space, TCP/IP (networking in general), disk defrag tools, terminal server, the entire office product suite...)
Hardware people don't design software; software people don't design hardware. So, neither is in a good position to help the other.
For actual board layout, yes, that's scripted. But the actual process of designing a circuit is not done with vi; it's done in the schematic designer GUI.
I guess you'd call me "very old school"... I prefer pencil and paper to any software tools. I can do things much faster and with a great deal less stress by hand. (with the exception of PCB layout. that takes forever by hand.)
Blocking the actual 911 call... I seriously doubt it. That's programmed into a phone switch that isn't connected to the internet (very, very few switches can be directly connected to the net.) Disruption to the E911 call center is a very likely. Someone will answer; they just won't be able to do anything.
Pay attention child. Go learn how spoofing works.
The 10/8 addresses are perfectly ok within the RR network. However, they should never be allowed to leave that network. In order for me to see those ICMP replies (TTL exceed, btw), they have to leave RR, pass through ATDN, to Sprint, to my provider, to me.
Pay attention to this part: A packet with a source address in a private (non-routed) IP range, passed through three (3) ISPs (2 of them teir1 providers) which means none of the them are lifting a finger to prevent spoofing.
It's not so much a matter of "knowing"... A lot of the stupid stuff you'll find (and if you actually go looking for it...) that is simply the result of "lazy" contractors.
As the saying goes, "Close enough for government work."
And just how does that differ from the way InterNIC used to manage domains? (Answer: it doesn't)
Obviously, no one read the part about an email address within NIPRnet. If you have an email address within NIPRnet, then you work for the DoD. It's not like someone can use foo@bar.com to register dozens of domains.
And an other thing, all the web engine does is fillout the templates that have been available via ftp for years.
(I'm assuming Google has already removed the reffered to pages from the cache.)
... because we all know the private address ranges are only unsed by end users behind NAT and firewalls.
...
Have you ran a traceroute to a Road Runner cable modem lately?
15 66.26.33.179 33.238 ms 32.589 ms 32.510 ms
16 24.25.1.101 34.965 ms 33.925 ms 34.576 ms
17 10.41.96.220 40.359 ms 40.125 ms 39.868 ms
18
(Hop 17 is a Cisco uBR925 owned by RR.)
The same was true for CAIS DSL service too -- they used the 172.16... range.