Community Rainbow Tables :: What Technology Next?

[cooper]

Remember that the USB key currently only works for Windows. Any *nix password types are pointless to generate untill the usb key works for *nix machines.

Which is unlikely to happen anytime soon since so far I've only seen windows market 'autoplay' as a feature rather than a bug.

It's interesting however to see if a Mac in its assumed userfriendliness works along with this... But let's not go there in this thread. Way, WAAAY off-topic.

[Shardz]

NTLM has just been given new appeal, (thanks mikesown) with a very interesting point made.

Silivrenion has imformed me that the largest rainbow tables for MD5 are 1.5TB, space that I doubt most of us have. I'm curious to know how large NTLM tables are (largest), and how feasible it is.

[linux4eva]

This kinda creeps me out. Couldn't a government create something like this and basically get into anything? You would need a lot of storage but....

[Shardz]

Of course. I'm quite certain they already do.

Just dont do anything that will provoke them. I'll see if the Marines has anything like it when I can.

[VaKo]

This kinda creeps me out. Couldn't a government create something like this and basically get into anything? You would need a lot of storage but....

We're doing this on old computers we found in dumpsters... They have data centers. Do the math.

(This is also why we need wiretapping and intrusion for evidence laws in place, to safeguard the public from goverment fishing trips and allow the law enforcement agencys to proactively deal with young men who are whiling to strap explosives to themselves, then walk into a mall. And child sex offenders.)

[Shardz]

I totally misread your post, and thought you were pro-government-wiretapping. (for those of you who saw this post before the edit, since it was an argument against that.)

I totally agree with you.

[VaKo]

I am in some cases, but what I'm against is goverments looking for crimes, any crimes they can find and use to bolster statistics, rather than having probaly cause to investigate a potential crime.

Its the difference between the cops smelling dope smoke coming from your house and then searching it, and the cops randomly dropping in to check if you have any dope.

[mikesown]

NTLM has just been given new appeal, (thanks mikesown) with a very interesting point made.

Silivrenion has imformed me that the largest rainbow tables for MD5 are 1.5TB, space that I doubt most of us have. I'm curious to know how large NTLM tables are (largest), and how feasible it is.

Unfortuantely, the bad news is that NTLM is about the same as MD5. The differences that make NTLM much harder to generate than LM are:

• - Password length. LM hashes are a maximum of 7 charactars, NTLM can be up to 100 or more. Generating from 1-9 or 1-10 takes up exponentially more space than 1-7.

- Hashing algorithm. LM uses a DES hash which takes a LOT less time to generate than the AES? hash of NTLM. Around 10x the time it takes to generate LM.

MD5 takes about the same amount of time and space to generate as NTLM does.

As for the problem of disk space, yes, most of us don't have 1.5TB lying around. The solution to this is using a distributed rainbow cracker. Give, say, 150gb to each person. If divided evenly among 30 people, the tables would provide complete access to the tables with 3x redundency. In addition, if someone was willing to volunteer a fast computer with fast drives(doesn't need to be high bandwidth, could be dialup) to host a webcracker, this would also solve the issue.

I believe that NTLM is the best choise by far simply because of the prevelance it will have in 1-2 years from now. If we start generating 1-10 all NTLM tables now, we could probably have them done within 6 months or so.

[Shardz]

Thanks a ton mike.

I'm not sure what to do next then... heh. Right now I'm trying to get my openbsd server up, to no avail. I've got alot of HD space I can donate (not give away, but have it being used for this).

[sleepwalker]

NT Lanman is the probably the best choice out of all of those.

[spektormax]

New Tehcnology Lanmanager v2 session would be the most useful because 1 there are no rainbow tables for it that are anythign above 95% for alphanumberic and they are already 7.5gigs. It be nice to do a full printable char (alphanumberic 32 symble and space) for a full 32 char? password. Essencaly though its jsut MD4 that the password is sotred int he %windir%system32config in the Security Acounts managnent (SAM) file. So technicly MD4 and NTLMv2 are the same thing in storage. Nt is transported thru HMAC MD5 so its network secure, but local sotrage is jsut an MD4. as for MD5,t hre are tables built, MD2 is rather unused, RiPMD is a rather expensive mathamatical rainbow table to reconstruct (the algarithum is much rmroe time consuming than MD5). Anything that is PUblic/Private Thats based on RSA (Prime numebr consept) is really impossible to construct a rainbow table since RSA keys are random as aposed to LM, NTLM and MD2,4,5. THerefore you would have to do a full 8-bit byte. SHA-1 is harder than MD5 (I dono how it compares to RIPMD). I havent dealt with CISCO. The fact is the most usful one for the comunity would be New Technology LanManager v2 session (NTLM that is curenlty used on XP SP2, Win 2000 SP4 and later)

[Matir]

MD5 is very widely used on nearly every major Linux distro. As mentioned, SHA1 is very popular in the Unix world and Silivrenion informs me that SHA1 is next on the list after MD5.

Do take note that Rainbow Tables are generally considered highly ineffective against UNIX-style MD5 or SHA1 passwords because those are salted passwords. See http://en.wikipedia.org/wiki/Rainbow_table..._rainbow_tables

[melco]

So when we begin to make new tables? I think MD5 and NTLM is the best choise.

If MD5 than 1-7all or 1-8mixalphanumeric?

P.S.

I think it will be good idea to store tables on several file hostings like Rapidshare.com or megaupload.com as fast mirrors, is'nt it?

[melco]

And additional question:

Can anybody rewrite program to use threads (for SMP).

I have 2 servers each witch 2*Xeon 5130. I want to make 2 tables and utilize all the cpu (4 cores) power, but now with current version of rtgen I have to run 4 instances of rtgen and make 4 tables simultaneously on each server :!:

[Jmtyra]

Hmmm, it looks like NTLM and MD5 are the choice of the community.

Should we try to tackle both, or would that be too much?

@melco, I have the same issue on a dual core box, and had to just do the same thing...run 2x/4x instances of the app. :(

Any-who, I'm game for either NTLM or MD5.

[Jmtyra]

As for the problem of disk space, yes, most of us don't have 1.5TB lying around. The solution to this is using a distributed rainbow cracker. Give, say, 150gb to each person.

I agree, very few [if any] have that much space available.

I'm seeding the Rainbow Tables, and have room for 150GB more for this project, but 1.5TB is not an option for me. :cry:

Of course, that brings up the question, who is going to build an interface that sits on a web-server that connects to 30x different boxes, of which will probably all have dynamic IP addresses. Even with services like DynDNS, it could be flaky at best.

[SomeoneE1se]

A few of us are working (slowly) on a Rainbow Cracker Web Interface and will be opening it up to the community some time in the future

[cooper]

And others are working (slowly) at making the cracking of the next thing a bit more like Folding@home. Big, central server, small blocks to be computed by clients, stats... the works.

[cooper]

http://www.hak5.org/wiki/Community_Rainbow...Assignment_List

Check that stuff out you guys!

Only 6 more tables and we're done. :D

[melco]

Why can't we start computing an other algorithm rather waiting 6 slow "amd-k6 or Pentium-mmx" for 1 year?

[cooper]

We could start immediately if you'd like. It just takes a 1.5 TB machine, a few nodes, some bandwidth and a lot of good will. And there's plenty of all of those to go around, isn't there?

[Heinzanova]

Space is cheap now... my closet samba server for media for my XBMC and other things is made up of 2.0TB as one massive LVM mount on 4x sATA2/300 drives... shitty little 2800 Speteron processor... but still that whole box was built for like $800 at Fry's....

and that is retail.

[cooper]

Great. Want to loan it out so we can fill it with the MD5 tables?

If you want to donate those $800 to the cause, that's acceptable too of course.

[Heinzanova]

Great. Want to loan it out so we can fill it with the MD5 tables?

If you want to donate those $800 to the cause, that's acceptable too of course.

So are you part of the MD5 team that was interviewed in Season1 Episode 1?

you know the EC2 cloud computer at Amazon might be something of interest... 10 cents an hour for some beafy server time use as many boxes as you need.

[mikesown]

Great. Want to loan it out so we can fill it with the MD5 tables?

If you want to donate those $800 to the cause, that's acceptable too of course.

So are you part of the MD5 team that was interviewed in Season1 Episode 1?

you know the EC2 cloud computer at Amazon might be something of interest... 10 cents an hour for some beafy server time use as many boxes as you need.

The EC2 cloud comptuer looks interesting, but I'd like to know how much power their cluster REALLY has. Even a cluster of 1000 computers would take quite a bit of time to compute all the NTLM tables(relatively speaking), probably like a week or two. Amazon charges by the CPU-hour, which would get VERY expensive as you add up all those hours.

All that being said, if the community was somehow able to amass a large sum of money($1,000? $10,000?), the tables could be generated on this service. Realistically, I don't think it will happen, but it would be cool if it did.