The mathematics that it takes to drive many of these algorithms was simply not practical in 1990.
although you young pups think we were using ROT13 back then, i assure you we had block ciphers, stream ciphers, differential cryptanalysis, and public key cryptography (dunno about related key and elliptic curve crypto). jeez DES is out of the early seventies, basically a derivative of IBM's Lucifer, a block cipher with an 128 bit key. you tried to brute force 128 bits lately? i assure you it would take one hell of a machine, and is beyond the accepted limits of serial computability. DES has been subjected to a lot of scrunity and analysis and the most effective attack is still brute force (the linear cryptanalysis / 40 bit brute force combo is not very practical!).
sure hardware solutions have improved, so we can higher effective keyspace without sacrificing performance during decryption and encryption, but that's a whole 'nother thread.
This document is serves more as a historical artifact now rather than a practial guide to decrypting like the government.
yes, this is no different than the stuff covered decades ago in books like Friedman's (elementary cryptoanalytics or whatever). i need to RTFD, but i doubt it even covers rotor systems like UNIX crypt and Enigma.
i'd recommend Applied Cryptography (perhaps partially superceded by Practical Cryptography) and Koblitz's Number Theory and Cryptography. they cover block ciphers, stream ciphers, discrete log, number theory, knapsack, elliptic curves, and plenty of other junk. for a historical or philosophical perspective maybe the following: Secrets and Lies, the Puzzle Palace, Codebreakers (unabridged of course). just some stuff i own and like.
if you are really "hardcore", i recommend a pilgrimage to Ft. Meade, Maryland, where you can visit the National Cryptologic Museum. it's great fun for paranoics and wannabe spooks.
30char password? Whats the point? I mean you can still brute force it
depends on the entropy. if the entropy is 7 bits per char (printables), then the effective security is 210 bits. no computer on the planet can brute force that, assuming the algo and implementation are decent. if you use an English phrase, the entropy might be 1.5 bits per char, and then it's doable.
i agree they are relatively insignificant, but interesting nevertheless. in the case of Honeynet Project, the sample size is too low. the probability of an attack will also vary with IP address allocation, perceived box worth, and other factors.
nevertheless an unpatched linux box is a bit safer than an unpatched windows box. with windows, you'll get nailed by DCOM, LSASS, or others via worms and autorooters. with linux, you will probably have a semi-manual attacker where there's a script kiddie at the keyboard enumerating and rooting the system.
the problem is one of time and money. a home user just can't be bothered to patch and harden his box. a net admin is already extremely busy. the hackers have time on their side, and by hiding behind enough zombies, they are relatively safe from prosecution. in rare cases, you'll have a skilled attacker who can bypass stackguards, chroot, IDS, and other safeguards. you can kiss you ass goodbye in those cases.
fortunately the more skilled attackers are often stealthy and inconspicious and at least have a set of ethics to which they adhere. the main problem today is the DDOS zombies, spam zombies and skript kiddies. this is one argument for partial or non disclosure of exploits.
there is very little interest in VMS among DefCon attendees. one needs relative familiarity with an OS to effectively attack it. and even if someone had a decent zero day, they are not going to waste it at a con. it would be either published for fame, or traded in the underground.
nevertheless, VMS is a strong OS. whole classes of buffer overflows don't even work. anyone interested in this OS should check out the Deathrow Cluster (featured on Slashdot) which invites non-malicious hack attempts.
disable unneeded crap in inetd.conf. add "--nolisten tcp" to X server args. for Windows, disable any unneeded services via start/run/services.msc. firewalling, decent passwords, and patches are obviously good too.
i wish more home users would take the time to do this.
the nifty thing about source code is you can split it up. the hackers don't need the whole source, because each segment has potential zero day. so instead of half now, half on delivery, you could split it up into $2,400 segments. using offshore accounts/servers and automated scripts, you could conduct this business very fast. in fact, there may be such systems already in place for software piracy.
certainly for $24,000 of profit, you could arrange an exchange in a country with lenient laws. tickets to EU / Caribbean are vanishingly cheap.
The only real reason to want the code is to find exploitable holes in the software.
certainly you can find holes without the source, but having the source is infinitely better. whoever bought it could probably be paid much more by Cisco to keep it hush hush.
whoever buys it will be spending loads of time analyzing it. might as well fork over an extra 24k for coders and exploit developers to pour over it.
i'm not sure how vulnerable PIX is to start with. but if it can generate 5 to 10 zero day (i doubt it), $24,000 is a bargain. blackhats hacking to steal source is decades old. Mitnick was trying for VMS source back in the eighties.
We also learnt that it's quite useful for you to have some kinda virtual machine like VMWare on your box, to boot into alternate OSes.
For even a home user VMware is great. Even supports running OS from physical disk in addition to virtual drive. For cost reasons, I'm going to evaluate XP and Virtual PC.
Anyone know if it's possible to use SecureIDs for your own personal home system (certainly at a price)
There are one time pass solutions like s/key and opie. They are bit more unwieldy than SecurID, but cheap. A challenge-response system would be cheaper than SecurID as well. I'd like to homebrew something like SecurID.
Yeah it's RSA SecurID, which uses an OTP (one time password) which changes every 45 to 60 seconds. AOL has used them for internals (employees) and overheads (contractors) in the past. They will likely use them to secure the host accounts, especially if hosts still have the permagag (chat gagger) ability.
A 6 digit SecurID would ideally take 500,000 attempts to brute force it. However, due to time windows (up to 10 min) and potential 45 second intervals, it can brute forced in ~ 30,000 tries worst case scenario. The passwords are one time typically which means you'd like need an exploit or brute force program that actually logs you in and hands off to the AOL client. Theoretically the main password can be acquired via a trojan or hybrid dictionary attack and verified. Then an attacker would brute force the remaining OTP of SecurID.
The AOL system is fascinating esp FDO91, CRIS, Merlin/Pegaworks, Kerebos, AOLserver, OpsSec, and Defender.
The Morris worm was way interesting. It was multi-platform as I recall, targetting Solaris and VMS. It used finger and sendmail exploits. Groundbreaking for its time. Here's an RFC about the worm: http://www.faqs.org/rfcs/rfc1135.html
I'll add that traditional viruses don't do well on linux since they are running at user level. They can't bind to the applications in/usr/bin, etc. So a traditional virus is almost pointless to use on linux.
That said, other forms of malware at non-root level persist. Ring 3 rootkits are useful to hide from a specific user, then once root is achieved, a ring 0 rootkit is installed.
He was probably referring to traditional viruses i.e. executable binders, which are rare. Linux has plenty of trojans and ring 0/3 rootkits available. Certainly some network worms exist on *nix but their propagation is thwarted by chrooting and lack of root privs.
There's also a GUI frontend available for Clam antivirus. However, Clam is not very popular in the various antivirus forums. http://www.clamwin.org/
There's an open antivirus project: http://www.openantivirus.org/
Several free antivirus programs exist like AVG, Avast, a-squared, and F-Prot.
I heard about the ease of Kryponite picking back in the mid-nineties. It was in the lockpicking FAQs. There's also an $150 pick that can open most of those barrel (?) type locks.
Home (non-institutional) MasterLocks were also easy prey before the 1998 versions. The last number could be determined in seconds and then the rest of numbers would fall into this formula:
n1 mod 4 = n2 mod 4 + 2 = n3 mod 4
This reduces 16,000 combinations to 100 (10*10*1) which can be brute forced.
Slashdot Mirror
every time you download music, God kills a kitten.
The mathematics that it takes to drive many of these algorithms was simply not practical in 1990.
although you young pups think we were using ROT13 back then, i assure you we had block ciphers, stream ciphers, differential cryptanalysis, and public key cryptography (dunno about related key and elliptic curve crypto). jeez DES is out of the early seventies, basically a derivative of IBM's Lucifer, a block cipher with an 128 bit key. you tried to brute force 128 bits lately? i assure you it would take one hell of a machine, and is beyond the accepted limits of serial computability. DES has been subjected to a lot of scrunity and analysis and the most effective attack is still brute force (the linear cryptanalysis / 40 bit brute force combo is not very practical!).
sure hardware solutions have improved, so we can higher effective keyspace without sacrificing performance during decryption and encryption, but that's a whole 'nother thread.
This document is serves more as a historical artifact now rather than a practial guide to decrypting like the government.
yes, this is no different than the stuff covered decades ago in books like Friedman's (elementary cryptoanalytics or whatever). i need to RTFD, but i doubt it even covers rotor systems like UNIX crypt and Enigma.
i'd recommend Applied Cryptography (perhaps partially superceded by Practical Cryptography) and Koblitz's Number Theory and Cryptography. they cover block ciphers, stream ciphers, discrete log, number theory, knapsack, elliptic curves, and plenty of other junk. for a historical or philosophical perspective maybe the following: Secrets and Lies, the Puzzle Palace, Codebreakers (unabridged of course). just some stuff i own and like.
if you are really "hardcore", i recommend a pilgrimage to Ft. Meade, Maryland, where you can visit the National Cryptologic Museum. it's great fun for paranoics and wannabe spooks.
is this available for Linux yet?
30char password? Whats the point? I mean you can still brute force it
depends on the entropy. if the entropy is 7 bits per char (printables), then the effective security is 210 bits. no computer on the planet can brute force that, assuming the algo and implementation are decent. if you use an English phrase, the entropy might be 1.5 bits per char, and then it's doable.
i agree they are relatively insignificant, but interesting nevertheless. in the case of Honeynet Project, the sample size is too low. the probability of an attack will also vary with IP address allocation, perceived box worth, and other factors. nevertheless an unpatched linux box is a bit safer than an unpatched windows box. with windows, you'll get nailed by DCOM, LSASS, or others via worms and autorooters. with linux, you will probably have a semi-manual attacker where there's a script kiddie at the keyboard enumerating and rooting the system. the problem is one of time and money. a home user just can't be bothered to patch and harden his box. a net admin is already extremely busy. the hackers have time on their side, and by hiding behind enough zombies, they are relatively safe from prosecution. in rare cases, you'll have a skilled attacker who can bypass stackguards, chroot, IDS, and other safeguards. you can kiss you ass goodbye in those cases. fortunately the more skilled attackers are often stealthy and inconspicious and at least have a set of ethics to which they adhere. the main problem today is the DDOS zombies, spam zombies and skript kiddies. this is one argument for partial or non disclosure of exploits.
there is very little interest in VMS among DefCon attendees. one needs relative familiarity with an OS to effectively attack it. and even if someone had a decent zero day, they are not going to waste it at a con. it would be either published for fame, or traded in the underground. nevertheless, VMS is a strong OS. whole classes of buffer overflows don't even work. anyone interested in this OS should check out the Deathrow Cluster (featured on Slashdot) which invites non-malicious hack attempts.
disable unneeded crap in inetd.conf. add "--nolisten tcp" to X server args. for Windows, disable any unneeded services via start/run/services.msc. firewalling, decent passwords, and patches are obviously good too. i wish more home users would take the time to do this.
so some idiot can go climb this
we are slashdotting microsoft.com
the nifty thing about source code is you can split it up. the hackers don't need the whole source, because each segment has potential zero day. so instead of half now, half on delivery, you could split it up into $2,400 segments. using offshore accounts/servers and automated scripts, you could conduct this business very fast. in fact, there may be such systems already in place for software piracy.
certainly for $24,000 of profit, you could arrange an exchange in a country with lenient laws. tickets to EU / Caribbean are vanishingly cheap.
The only real reason to want the code is to find exploitable holes in the software. certainly you can find holes without the source, but having the source is infinitely better. whoever bought it could probably be paid much more by Cisco to keep it hush hush. whoever buys it will be spending loads of time analyzing it. might as well fork over an extra 24k for coders and exploit developers to pour over it. i'm not sure how vulnerable PIX is to start with. but if it can generate 5 to 10 zero day (i doubt it), $24,000 is a bargain. blackhats hacking to steal source is decades old. Mitnick was trying for VMS source back in the eighties.
We also learnt that it's quite useful for you to have some kinda virtual machine like VMWare on your box, to boot into alternate OSes. For even a home user VMware is great. Even supports running OS from physical disk in addition to virtual drive. For cost reasons, I'm going to evaluate XP and Virtual PC.
It's named after a Paul Dirac, a British scientist who worked on quantum mechanics.
In this case, your couputer would use as much power in 10 minutes as it did powering up. "as much energy" is correct. Energy is power over time.
Anyone know if it's possible to use SecureIDs for your own personal home system (certainly at a price)
There are one time pass solutions like s/key and opie. They are bit more unwieldy than SecurID, but cheap. A challenge-response system would be cheaper than SecurID as well. I'd like to homebrew something like SecurID.
Yeah it's RSA SecurID, which uses an OTP (one time password) which changes every 45 to 60 seconds. AOL has used them for internals (employees) and overheads (contractors) in the past. They will likely use them to secure the host accounts, especially if hosts still have the permagag (chat gagger) ability.
A 6 digit SecurID would ideally take 500,000 attempts to brute force it. However, due to time windows (up to 10 min) and potential 45 second intervals, it can brute forced in ~ 30,000 tries worst case scenario. The passwords are one time typically which means you'd like need an exploit or brute force program that actually logs you in and hands off to the AOL client. Theoretically the main password can be acquired via a trojan or hybrid dictionary attack and verified. Then an attacker would brute force the remaining OTP of SecurID.
The AOL system is fascinating esp FDO91, CRIS, Merlin/Pegaworks, Kerebos, AOLserver, OpsSec, and Defender.
The Morris worm was way interesting. It was multi-platform as I recall, targetting Solaris and VMS. It used finger and sendmail exploits. Groundbreaking for its time. Here's an RFC about the worm:
http://www.faqs.org/rfcs/rfc1135.html
I'll add that traditional viruses don't do well on linux since they are running at user level. They can't bind to the applications in /usr/bin, etc. So a traditional virus is almost pointless to use on linux.
That said, other forms of malware at non-root level persist. Ring 3 rootkits are useful to hide from a specific user, then once root is achieved, a ring 0 rootkit is installed.
So, why don't you stop spreading FUD?
He was probably referring to traditional viruses i.e. executable binders, which are rare. Linux has plenty of trojans and ring 0/3 rootkits available. Certainly some network worms exist on *nix but their propagation is thwarted by chrooting and lack of root privs.
There's also a GUI frontend available for Clam antivirus. However, Clam is not very popular in the various antivirus forums.
http://www.clamwin.org/
There's an open antivirus project:
http://www.openantivirus.org/
Several free antivirus programs exist like AVG, Avast, a-squared, and F-Prot.
My AOL cracker needs a plaintext file!
Dunno if this works against Kryptonite, but here's a tubular lock pick:p roduct=48
http://www.lock-depot.com/Scripts/prodView.asp?id
I heard about the ease of Kryponite picking back in the mid-nineties. It was in the lockpicking FAQs. There's also an $150 pick that can open most of those barrel (?) type locks. Home (non-institutional) MasterLocks were also easy prey before the 1998 versions. The last number could be determined in seconds and then the rest of numbers would fall into this formula: n1 mod 4 = n2 mod 4 + 2 = n3 mod 4 This reduces 16,000 combinations to 100 (10*10*1) which can be brute forced.