hpr-knowledge-base/hpr_transcripts/hpr0215.txt

Episode: 215
Title: HPR0215: Guide to using linux Rainbow tables
Source: https://hub.hackerpublicradio.org/ccdn.php?filename=/eps/hpr0215/hpr0215.mp3
Transcribed: 2025-10-07 14:03:36

---

music
Well, hello and welcome Hacker Public Radio listeners to Phoenix's Student Huckers
Guide to Linux.
My name's Aaron, but you guys as usual can call me Phoenix.
Today I'm going to be talking about Rainbow Tables and how they can be used to crack passwords.
As usual my podcast is for educational purposes only and it's not here to teach you how to
hack people.
A little bit about myself, my name's Aaron Finnan, sometimes known as Phoenix, I'm an
ethical hacking student at Avatar University Don D. I've been doing this for about 2-3 years.
I'm also a very passionate Linux user and I've been the current president of the Avatar
Linux Society for about 3 years now.
My show notes will be available on the Linux Society website which is www.thelinuxsociety.org.uk
and that's www.thelinuxsociety.org.uk and as usual my slides will also be made available.
My show notes will also be made available at our good friends over at the Linux Basement
which is www.linuxbasement.com.
The reason that I wanted to talk to you about Rainbow Tables is to actually find this stuff
really interesting and every geek likes to know something about something and this is
quite interesting stuff.
I mean I own up it's as far from Linux as I could get, apart from the prime lesson
here is that Unix is not vulnerable to these sort of attacks.
This is to do with sorting your hashes and all of this stuff which I'll explain at the
end when I talk about possible countermeasures.
Although I say that this is as far from Linux as you could possibly get, I do use Ubuntu
with a package called offercrack for cracking or lag manager hashes and my examples in my
show notes will be using that as well.
offercrack is available in Ubuntu repositories but I think it's available in the restricted
ones.
It's also worth knowing about this type of technology because WPA is also vulnerable
to this sort of attack as well but not as much.
What I really want to talk about is it puts the reality of what would seem impossible
possible.
You can take 7 trillion possible password combinations in minutes with a 99.9% chance for probability
when using Rainbow Tables and that's just mind blowing stuff when you think about it.
I mean why would this stuff be of interest to you just in case because I said so isn't
enough.
Personal security.
I promise that your views on password security will change hopefully after this talk.
If you've not already adapted strong password security this should scare you enough to do
it.
I haven't met anyone that isn't surprised by the power that Rainbow Tables harnesses.
I'll also show that this technology is available to everyone out there.
Fingers crossed you'll listen to this, go home and try it and it will work with ease and
what that will show is that in reality anyone can do this, anyone that can use Google can
password crap using Rainbow Tables.
Like I say one of the tools that I use is available in the Ubuntu repository so it goes
to show you how easy it is to get a hold of.
This could be used against multiple organizations that use land manager hashes and you probably
have contact with them so all that you might not be vulnerable people that you know could
quite possibly be.
And if I say it and as I've said before if you haven't I hope by the end of this podcast
that you'll start to implement a strong password policy.
So what I'm going to kind of go through today is a little bit of terminology, talk about
the mistake that land manager hashes made and what they're about and how that mistake
is sort of shown into the fabric of time.
I'm going to talk about how you use Rainbow Tables, I'm going to touch a little bit of
the science about it, I'm going to talk about some of the countermeasures and also how
Rainbow Tables are not just for windows hashes and other types of attacks that can use for
this as well.
Now terminology is not my strong part and I'm not much of a mathematician either so you
have to bear with me but when I talk about a hashes what I'm talking about is a hashes
kind of like a short identify a fingerprint or something more complicated.
Hashes are produced using a one way mathematical function they're used in like database systems
and security and cryptographic systems as well.
They used to store passwords.
I mean in English they're unique like your fingerprint and they can't be reversed like
you know like 1 plus 1 is 2 you can't you know you can reverse that mathematical function
you can't reverse a hashes.
It's been used in security for a long time but it's also been a way of verifying data
integrity as well.
I'm sure plenty of people have downloaded distributions and you'll see that there's a check
some MD5 hash that you can calculate to make sure that your disk that you're downloading
has not been corrupted or hacked or something like that as well.
So you could you know the author puts up what the the hashes and then you can check that
across.
I mean I'll talk about how we use that in passwords and in operating systems.
I'm also going to talk about lookup tables and computer science lookup table is a data
structure usually in a ray or an associative array and often used to replace run time
computations.
It just means that there's less processing time being used at the time and the memory
is easy to the answer is easy to pull out of the memory once you know this lookup table
is loaded.
I'm also going to talk about reduction function.
It's not easy to explain well it's not easy for me to explain is probably the fair
apart.
I mean I'll try but you can't understand understanding a reduction function is not
pinnacle to today's episode whatsoever at all.
In essence it's a way of solving a complex problem by turning into a simpler problem.
I either sounds like a making sense or it sounds like I've been smirking something to
be honest with you but the reality of it is is a reduction function is a way of changing
one complex problem into another simpler problem and if you keep on reducing the problem
you eventually you'll find a way to solve it you know like I say allow the make sense
to you or the one but as I say it's not pinnacle that you understand this to being able to
use rainbow tables so don't panic if I've done enough job of explaining it I'm sure
there'll be plenty of people that know kind of a little bit more about it.
Now what I really want to do is set kind of set the scene I mean you have to remember
in the beginning there was a world without passwords and computers were a resource that
was shared that weren't cheap there were big mainframes and people went and spent time
in front of it then eventually someone decided that they needed to protect their system
from unwanted users and you've guessed that passwords were implemented and as the word
says you know password if you get the word right you pass through and that's the it's
not really rocket science a good old friend Stolman had you know it has a famous talk you
know I'm sure you'll be able to find it online about how he was at the artificial intelligence
lab at MIT and they implemented passwords and he had the password database and it's very
interesting I'm sure you'll find it on Google video or something like that.
The problem with locks is that they have to be unlocked with a key or if they have to
be unlocked into a key they can be tricked into thinking they're being unlocked with the
right key you know lock picks tend to trick a lock depending on the skill of the picker
you know and in in reality you lock a door humans always want to know what is behind it
well in the computer terms it's a sequence of characters that can be course referenced
with a particularly user to verify their identity that's the lock and that's the key the
lock is the system saying no the key is your sequence of characters that the computer
knows if you tell it that that it should let you through as that user with everything
that's tangible you know it can be recreated or simulated in computing you know much so
what we can do is we can simulate the sequence of characters and that will let us through
the door it's not really you know like I say it's not really rocket science I mean the problem
lies that in the reality you need to store the character sequence on the computer so it can be
cross referenced against the user now basically what we're saying the plans for the key are on the
system does someone could just look at this this character sequence and say oh I know your password
not so bad if it was your family but if you're in an organization and you want people to use
individual accounts you know what people to be able to look up other people's passwords so we
need to find so so there was a we need to kind of find a way of storing the plans to this key
onto the system without it being understandable to the user what I mean by that is that we can't go
and look it up and say oh I know your password is password and we need to be able to store it
on a system so that if anyone looks it up it doesn't make any sense at all and this is kind
of where hashes come into to to their fruition I mean a hash can't reverse this store the hash
when a user inputs their password it's converted into the hash and if the two hashes match
then then they come and if they don't you keep on retrying pretty clever if you think about it
because what this means is the computer never stores the password on the system one of the
computer does is store the hash so you input your password it puts your password through the
hashing function and if the two hashes match in you go if they don't you don't get in so it's
very very simple but it's very very clever because like you said before we don't have to store
the actual password on the system but the problem really lies is that humans have a problem
of remembering complex things so we all tend to you know people tend to use simplistic passwords
you know stuff that are words that are easy to remember and in reality you know you
a system that relies on a human being is always vulnerable to the human factor and something to
remember I mean you know the problem really lies in here that with enough force locks will always
be broken so we can if we go back to remembering this human factor of not being able to not storing
you know using passwords that are easy to get a part of a dictionary we can start
brute folding the lock until it eventually opens it doesn't have you know and that is basically
the first way the first kind of a car to use against breaking password security was just to basically
make a list of possible passwords basically a dictionary and keep on trying those possible
combinations until the password hashes the right one two hashes match and in you go
the more complex the harder it is for people to to remember so humans tend to choose the path
of least resistance so they always choose something easy to remember and if it's easy to remember
then it's even easier for us to guess and that's something to remember as well
so as you can imagine the countermeasure really to this was to start storing passwords that
were not in the dictionary out for new merits so password one is not going to be an
dictionary it's still very guessable but if you start separating words with letters and replacing
characters with special characters what you'll find is they're not an dictionary they don't become
easy to guess and it takes more and more and more time to stop brute forcing this lock
um so as I say password policy started implementing that you need to use
often new merits and and and whatnot we've all seen them before so we go back to how
these are stored on the system and I spoke earlier on about land manager hashes or LM hashes for
short um this is a mistake that's pretty much been sewn into the fabric of time and as ideas go
this is probably this is probably a stinker of all ones our good friends at Richmond think
that we lovingly call Microsoft came up with this in fairness I don't think anyone could have
guessed the knock on effects that this mistake would have had but I'll explain what a land manager
hashes well I explain how you get a land manager hash how Microsoft do it to take a password
and then converted all often into all uppercase characters um so if you know you had a capital P
a lowercase A a capital S a lowercase S a capital W a lowcase O a capital R a lowercase D
and the password would become password all in uppercase if it's longer than seven characters they'll
divide it into two sections containing seven characters so if we took the word password the first
would be P a S S W O R then they would get another they would get they would split into another
plate containing the D and the rest would be the folder space then the hash each part individually
and then put the two together and they call that land manager hash it's a very simple explanation
I'm sure that some of you guys you know we'll think that's a really layman's version but it in
essence is is how it's done so what's so wrong with doing it that way well first converting
everything into uppercase cuts the possible combinations down massively by a by a factor of 26
also the hash for an uppercase password and the hash for a lowercase password will generate
two different two different hashers so by capitalizing everything we knock out a factor of 26
off the possible character set so if we had the whole possible character set including capital letters
there would be 95 keys and most keyboards to choose from by uppercasing everything you cut that down
269 splitting the hash splitting the password into two has the following sort of consequences
69 to the power of seven is a hell of a lot less than 69 to the power of 14
69 to the power of seven works out about seven trillion possible combinations by
in comparison 69 to the power of 14 is about 55 set trillion possible combinations and trust me
that is a real word set trillion it's s-e-p-t-i-l-l-i-o-n as I say we're able to attack both halves
independently which means that we are only ever cracking a seven character password
regardless of the length of the password the character if the password is you know 14 characters
we're still cracking two seven character passwords and as I say they can be cracked independently
but however don't get me wrong seven trillion is still a really big number
um but splitting it up into two bits makes the process really simple for us and in the reality
this has been this mistake has been etched in time for backwards compatibility
all Windows systems have the ability to store the password as LAN manager hashes however in
Vista this option is turned off by default and in exputes turned on by default so they are trying
to phase out but it's going to take a long long time for this to be to be done I mean the idea
of the tradeoff is really simple if we have enough time we could generate all the possible
hashes of up to seven characters we could store them and use them time and time and time again
okay it takes very little memory to brute force each seven character combination however
it takes a lot of time to do it if we store all of these possible character combinations
these seven trillion possible seven character passwords and we say each password
takes up about 21 bytes of data okay if we multiply that by about seven trillion I'll give you a
rough idea of the amount of space required to store it is somewhere in the region of 145
thousand terabytes of storage you would need which is still pretty unreachable for most people
in today's standards now but there is an answer and this is using the time memory tradeoff we do
some of the hard work first we do a lot of the mathematics at the beginning in essence we trade
time memory in space and we find a happy medium so I mean why I was you know what if I told you
there's a way that we could cheat here that we could store all seven trillion possible password
combinations without actually having to store the physical seven trillion possible combinations
and this is where this reduction function that I talked about earlier on comes into essence
we store the start like the clean text passwords at the start we put it through a reduction
function seven times and we store the final output and then we just delete everything in the middle
and what happens with this is that what a rainbow table in essence does is takes the hash looks at
the end of this chain and if it finds a match then it knows what the password is if it doesn't find
a match at the end of this chain then it uses a reduction function and the hash moves along the
column until it finds the right possible combination it just means that we're able to store
huge amounts of possible data without actually physically storing the data you know and if we find
that it's going to be in that chain then we can just unfold that whole chain and find the answer
and that in essence is what a rainbow really does and we store a lot less data and we do a lot
more into computations at the beginning but it means that we have this data set that we can use
time and time and time again I mean like I say my sure notes will have a little bit about the time
memory trade off if you want to find out anymore obviously a quick google and you'll find lots of
more information there as well so in essence you know what is a rainbow table and well it's basically
imagine this massive lock up table full of all the possible combinations lots of chains side by
side like as you cut the middle out and then you store the beginning and the end of the chain okay
so you're left with the answer and the last computation in the chain from that if it's not in there
we can work back the chain and use some of the processing time at the time use some of the memory
that we have on the system and work out the password if it's likely to be in that chain so we're
not having to generate thousands and thousands and thousands of possible wasted processes cycles
working on something that's not going to be an answer here we're limiting our search to
that physical set there I mean in essence it sounds pretty simple but you know it takes a
you know it takes a while to it takes a while to generate these tables it was a guy called Philip
I think his name is Philip Oshlin and Terrible it pronounced since I named him so I hope he doesn't
take any offense but he kind of came up with this idea and he went and worked on the offer crack
kind of software which made this all you know available now what I did and I'll make these notes
available as I said I downloaded a copy of offer crack from the Ubuntu repository and then I also
downloaded some hash files now there's lots of software that you can use to capture land manager
hash files I think it's PW, PW dumper in the program is called but you can basically get the
where a hash is normally saved on a Windows system I think it's at system 32 config and I think
it's a SAM file as A.M I've talked about that the last time and I did changing Windows passwords
as I say I mean I installed a copy of offer crack on my laptop and from the same website I was
able to download a set of rainbow tables that had a 99.9 percent chance of cracking these passwords
I was able to put in this 32 these 32 passes that I had got off this site for testing purposes
they were made available I'm sure you can find them online they'll be available in my notes as well
and with this table I was able to crack nearly all of the passwords on a low spec system in about 10
minutes it was pretty phenomenal stuff that size of the table was only 710 megabytes so you know
it's pretty easy to put the size of a distribution when we think about it very easy to do now to
basically have all the possible passwords all the possible character sets the actual rainbow table
size is only 64 gigabytes so basically to crack nearly any LAN manager hash which is pretty much
any Windows system bar Vista you'd be able to download a set of tables that are 64 you know 64
gigabytes and size it I mean it's in this day and age you can buy a terabyte hard drive for 99
pounds and you know so it's not not unavailable to everyone the the beauty about using Linux is
is that we we're not susceptible to this sort of attack we use something called salt and your
hash so even if we had the LAN manager hash thing which we don't but even then we would take
this password and then we would use what's called a salt which is something on our system unique to
our system okay to generate our hashers from which would mean if we were needing to generate
rainbow tables to hack Linux systems what we would need to do is find the salt of the system what
the system is using to salt and then generate all those tables now the 64 gigabyte data set that
I was talking earlier on the rainbow tables takes two years to compile okay so if you had to do
that every time you wanted to hack a Linux system you can imagine it's just it's not worth doing
and hopefully someone would have changed the password by them you know nearly all other operating
systems actually employ something like this which is apart from Microsoft they don't seem to
salt their hashers at all which is very strange but stuff like you know open BSD does Linux does
I think open BSD salts its hashes with I think 128 bits salt which you know gives you a massive
idea about how long it would take to to generate these tables now some of the possible other
countermeasures to this as well is that if you think that the time it takes to compute these tables
kind of goes up dramatically the minute that you pass 14 characters the time it's just a huge
amount of time so if you employ a password that's longer than 14 characters it's a good start
you see in a Windows system if you use a password that's longer than 14 characters it will have
to be stored as an NTLM hash which doesn't get split which doesn't get capitalized it means that
you're not going to be susceptible to these like to to this rainbow attack but when we we talked
about password security at university one of the things that we kind of really
bang the drum about is actually pass phrases are a lot better than password lots of reasons
a pass phrase will generally take you past 14 characters but it also makes it very simple to
remember as well because if you think about spaces spaces are a character as well so if you use
something like my past you know my space bank space password space is space this this will take
you through huge amount of possible combinations compared to if you had a password of only 14 characters
or something like that so pass phrases are a far more secure way of securing something they're
also incredibly easy to remember so you know you could use you know people as I said earlier on
humans tend to take the path of less least resistance so say you're going to choose a password
that's based on your your favorite football team for me it's going to be very easy for me to sit
back and say all right that guy supports ranges so I'm going to try ranges one ranges two ranges three
ranges four blah blah blah blah blah now what happens with a pass phrase is we could do something
like the sentence my favorite football team in the world is ranges it's going to be incredibly hard
to guess and it's going to be incredibly hard to sell rainbow tables so you're going to need
you know petabytes of data to store the possible sequence in that so you know one of one of the
good count measures here is to remember that that using a pass phrase is easy to remember brings out
huge amount of characters involved it's easy to type I mean that's the requirement for a good
password really easy to remember long and easy to type well that's why a pass phrase is probably
better than a password there are other uses for rainbow tables as well they can use they can be used
to crack lots of different tables as well there is a there is rainbow tables for the ntlm
hashes the md5 hashes the sh1 hashes but they can also be used for hacking wpa psk as well
not to the same extent because when what happens for for for the hashes for that is that they are
generated using the ssid as well so you know you need to know the ssid then you need to generate
the rainbow tables and they're each generated off each ssid now there is a list of the thousand
most popular ssid's and pound to a penny belkin 45g will be up on the top there and so on and
so forth but as I say it's kind of a type of salt so if you use a very individual ssid for
your wireless router chances are you're not going to be vulnerable to a wpa to a rainbow attack
on your wpa I think the data set for a wpa rainbow table is about 500 gigabytes so it goes to show
you you know you need a lot of bandwidth now the other thing to remember with rainbow tables as
well and I apologize because it just dawned on me that I haven't mentioned it before most of
these rainbow tables are available for download you don't need to generate 64 gigabytes of data
you can actually download them there's also a lot of websites online link free rainbow tables.com
that have you know massive massive massive rainbow tables that you can download via torrent
that you don't need to pay for or anything like that so that's always worth remembering as well
I mean every like I say every possible character set for the lm hash total 64 gigabytes I mean it's
you know so within the reach of everyone is unbelievable the software we use to generate these
tables and to use these tables and to store these tables is for most parts free and available on
the internet as well and as I say I'll have lots of links available in my show notes so if that
you if you are interested please come along for Linux to say any dog dog UK and have a look at
the slides and the show notes that I've got for this so kind of enclosing Google is your friend when
it comes to searching if you find an MD5 hash why don't you put it into Google it sounds really
strange but Google is this in reality this massive storage engine out there it's not just a
search engine if you take like the MD5 hash for phuba and stick that into Google it will tell you
that it's an MD5 hash for phuba not all of them will be in there but I think there's a story about
how how a website was hacked and then did the guy can understand how he done it you know he
managed to get the MD5 hash but it just didn't work out how he worked out what the password was
and in the end what happened is the guy taken the MD5 hash and stuck it into to Google and you know
someone had already cracked it and there was the password took seconds to do but it sounds stupid
but I have to say this to everybody every day you know lots of people every day when it comes to
finding out stuff use Google you're going to find the answers there the information has been out
there for a long one time on this this I think this technology is 10 15 years old
passwords are always weak regardless of the hash and algorithm if you use something that's
simple to remember or simple to guess then you're always going to regardless of what's used
you're always going to be vulnerable to someone going I guess your password is so remember try
and use phrases they're very very hard for people to guess they're incredibly hard to generate
tables for and should kind of take you off this you know take you off this attack remember you
should always try and make your password over 14 characters long it makes the computation time
to generating tables virtually you know virtually impossible you know when you start using 18 and
19 character passwords they're incredibly hard to guess they're incredibly hard to generate tables
for as well and as I say never let Microsoft choose security for you your job to choose your
securities your job to investigate what's the best for you as I say all of them I know it's
about how you can go about using rainbow tables and some of the some of the stuff that I've got
will all be available on my on our Linux Society website and also on the Linux basement as well
so hopefully you're going to have enjoyed this episode today and go out and play
and use this technology now remember don't steal someone's hashes and crack them you'll be
breaking the law if you do that and you could face the jail okay so that's Phoenix Student Hacker's
guide coming to an end for this month before I go I would like to do a big shout out to my friend Chad
Wallenberg over at the Linux basement he'll be coming up to episode number 3030 happy birthday
over there you guys Chad you've done a fantastic job and let's hope the next 30 is just as good
as okay anyway thanks very much this has been Phoenix Student Hacker's Guide to Linux this is
been exciting off
thank you for listening to Hacker Public Radio
HBR is sponsored by caro.net so head on over to C-A-R-O dot anything for all of us