Think Complex Passwords Will Save You?

let's get started and welcome David thank you all right everybody hear me okay not sure if they may be a little bit testing testing hello well try this out okay so my name is David Holton and I'm gonna be talking to you about a few different techniques through the service that we put together as part of the tour an organization that I'm a part of for attacking specifically the DES encryption algorithm and attacking that versus the password in order to try to I guess break into systems without having to guess complex passwords so just a show of hands here did anybody catch the talk that I gave in 2012 with Moxie Marlinspike at Def Con

a couple of people okay well so I'm just gonna do a really quick recap of the talk just because there's a little bit of background with this and the service that we provide so in 2012 give it give this talk where he demonstrated a hundred percent break of mschap v2 which is a mutual authentication protocol that's used with pptv VPNs and WPA enterprise and it's basically a mutual authentication protocol that allows you to use a password to authenticate to a network and this is nothing new this protocol has been around for since the late 90s and and since late 90s this paper specifically came out that showed that breaking Amos chapter e2 was just a

matter of breaking a few DES those operations which is around 2 to the 57 and and so it was well known for a really long time that you know obviously state actors well-funded organizations could perform this attack but still people use pptv VPNs and this was integrated into WPA enterprise and and for some reason this kind of still gained traction as a standard so a quick overview of what mschap v2 looks like you basically have you know a client that says hello to the server the server sends back a server challenge the client generates its own challenge and it kind of you know mixes it together to create this challenge hash and then you basically take your NT hash which is

just an MD for of the users password and then that's used to encrypt this challenge hash and creates a response sends it back and that's how you verify that the client has the same password as a server and so the stuff that's sent over the network you can obviously see the cipher text and and then also that's known plaintext is sent across as well so we basically have everything except for the key that you know - there so so we have all the pieces that we need in order to try to brute force these these keys and so up until our talk people mostly focused on attacking the user password so doing dictionary attacks trying all the low-hanging fruit of easy

to guess passwords and and then using that with the whole authentication through programs like asleep to find a password that would work for cracking cracking this network and if you look at if somebody had most complex possible password for for their user password it ends up being a lot of characters there you know a very large key space to search around 92 bits and so we were wanting to come up with an attack that would work 100% of the time no matter how complex your password was so then T hash in a lot of these cases is essentially a password equivalent a lot of these networks will let you authenticate directly just with an NT

hash you don't have to have the user password and so we specifically worked on trying to crack the NT hash and so that meant cracking just the des opérations so what they outlined in the paper was basically there's these three des opérations that all have the same plaintext all these are just encrypting the challenge hash and all that is the same and and then T hash is actually 16 bytes long so in this last one there's only two bytes of an actual key that's being used for the encryption operation so that one you can use do with an EPC out there you know it's just 64,000 possibilities these other two required you know full des

brute-force and so that's really the the issue with trying to crack this empty hash from the from the authentication that you capture so we have the challenge hash and and so a naive implementation of doing this would be you would you know just do a brute force on on both of these keys and you would do you know for the first key and then you do it for the second key and be a total of you know around to the fifty-seven operations you have to do but because we're dealing with the same plain text in both these cases because both are encrypting that one challenge you can actually combine these together and just do two compares after the after

the decrypt and are they encrypt and then so you only really have to go through the key space once just with two compares so we we decided to okay this is this is a great attack now we just need to be able to brute force DES and and we also wanted to let everybody be able to do this and make it you know something where people can't ignore this anymore this is something that anybody is able to do so Moxie put together this program called chap crack where you can capture a PPTP VPN authentication run it through this tool and it'll extract out all the known plaintext and cipher text and create a token that you could submit to his

website that he had up called cloud cracker and and then you know within 24 hours ideally we would be able to send you that NT hash that we cracked and so how easy is des to crack these days back in 1999 the FF built a DES cracker with a bunch of custom Asics and a custom about a quarter million dollars and there was their system took around nine point two days worst case time to crack a DES key and we wanted to do this in 24 hours and so looking at all the possible options out there just looking at like AWS using CPU cores to do something like this would require around 80,000 CPU cores

working for 24 hours which would cost around $125,000 with AWS p1 instances which is their high end GPU instances it would take around 1800 GPUs and that would be around $20,000 which is still very expensive and my background is in working with FPGAs and so I just happen to have a bunch of FPGA hardware laying around for my company and decided okay well we can take all these all this hardware that's you know just going end-of-life right now put it up online and crack des with this and with basically one for you system we can crack des and around 24 hours and so let's just put that online provide the service and we'll just charge like $20

to do this because we didn't want everybody just flooding the system with with requests and and then that ended up moving into my basement so this is my little room in my basement with with the server in it and so this is basically the the guts behind a cloud cracker and our new service called crack de Sh so then of course you know this is vulnerable everybody can crack des now so then of course everybody rushed to fix everything right right right right so we specifically called out I predator in our talk in 28th 2012 and I went and looked at their webpage recently and they just put a little warning on there like oh by the way PB TP is kind of

insecure so you may want to use it we still of course support it so everybody's still connect with it but just a little warning and this is pretty much most of the VPN providers out there just put up a little warning and said oh just keep using it it's fine and then also with WP Enterprise people just kind of said oh well of course everybody does strong certificate checking in the enterprise so that's not really an issue either so it just kind of got dismissed out of hand and so we started running the service and we were doing quite a few jobs and we started noticing some very interesting jobs come in like this one

for example where it's a plain text one 1 two 2 three 3 four 4 that doesn't look very random to me does anybody have a guess of what that might be okay and then then this one where both of the cipher texts are the same value so so with that it seems like they're not doing Emma's chappy - they're just trying to crack a des key and then I started seeing some articles you know people actually using the service for doing for cracking WPA enterprise and the one thing that caught my eye related to to this where we see the one 1 two 2 three 3 four 4 is that the default challenge that is sent out with SMB capture module

with Metasploit is one 1 two 2 three 3 four 4 and and their articles coming out saying that you can basically use the service to crack samba authentication like land man and ntlm v1 authentication with Windows so people were obviously using the system for more than just mschap v2 and and then one day traffic just kind of dropped off I started stopped getting jobs to my system from Cloud cracker and from Cloud cracker comm mysteriously disappeared I have no idea what happened and I messaged moxie and everything and he never applied to anything so I you know people were emailing me saying hey they need to run jobs on it and so I ended up creating

this new site called crack SH which is run under the tour account organization and it's basically just specifically dedicated to DES cracking and it works pretty much the same it takes all the same tokens that that cloud cracker used to do for Framus cheffy - so while I was going through this process of kind of reinventing the service I did some research and I was trying to figure out what people are actually using the service for so we could expand out the different things that we support and and one of their features they might want to add in and the real goal of this whole project from the beginning was how can we kill DES once and for all you know

that it should have been killed back in 1999 when it was demonstrated that people could could break this easily and now you know almost ten years later it's still out there pretty prevalently so how can we make sure people finally switch away from this so um looked into the windows authentication a bit more and it's pretty much I think that windows authentication technically uses Emma's chap v1 which is very similar to Emma's chap v2 but the challenge is provided entirely by the server instead of hashing it with them you know the server and the client challenge and so it actually makes it makes it a bit easier and that's why you can actually provide your own challenge to the clients that

are connecting of you know one one two two three three so so they're they're articles online on how to use this and then obviously people were using this would be capture module for this and and basically the values spit out from this imbue capture module you can plug directly into cracked-out SH and then within 24 hours it'll spit out the NT hash that you can use to authenticate to the system and then also same with if it has anybody here use responder responders another popular one that basically kind of replies to any sort of NetBIOS resolution requests and and then you connect to this fake server and they can do downgrade attacks and things like

that so if you tell it to downgrade you can get clients to connect and try to negotiate land man and ntlm v1 and they'll actually just kind of automatically start doing negotiation with the current user that's logged in so you can start grabbing their you know their credentials and then those values can play plug directly into cracked-out SH and then you can get then T hash for those users so I also looked into WPA enterprise and back when we originally did this research people are using free radius W PE which is kind of a patch for free radius that just spits out the challenge and response parameters but now there's this other program called hostapd w PE which is a lot more popular

and and so we're going to be doing a demo actually a def con with with this and have some more details about what clients actually do strong certificate checking there and also at our rocket shot at a couple weeks so so now we ended up adding in the support for basically taking that these net ntlm hashes that you can normally feed to John the Ripper and so now you can put those directly into the system and then that will crack the NT hash for WP enterprise and then once you have this NT hash if you want to actually connect to the network you can use WPA supplicant or there's you know various clients that'll let you use the direct

NT hash instead of the password to connect to the WPA Network another thing that that we saw right after our talk in 2012 was Karsten Knoll gave this talk on reading SIM cards basically sending over their updates to SIM cards and a lot of these are secured with with single DES I think a lot of the newer NZ's Triple DES but he actually basically used I'm not entirely sure if you use dart service or not but but his theoretical attack you know called for having something that could crack DES in order to crack these over there update keys so you can send remote updates to SIM cards to listen in on their phone calls and read out their

contact book and you know there's a full Java Virtual Machine in there that you can you know rewrite and put them back doors and stuff like that so and then also if you guys watch mr. mr. robot this was on one of the episodes so that was kind of cool and then we also wanted to provide just a general-purpose interface for people to be able to crack DES no matter what sort of protocol they're looking at and so we put together an API and we wanted to have some really simple rules so we basically just you know say you can specify a cipher text and a plaintext and a mask for the compare and you can say whether

you're doing an encryption or decryption and then as long as there are I think we limit it to you can only have like 24 0 bits and in your mask and so as long as you're within that then we'll send you a list of all the possible keys that match that and then you can use software to quickly go through all the candidate keys if you're you know wanted to compute some checksums or do some others thing to verify if you found the right here or not and so we put up this whole interface and and then we wanted to figure out what kind of an application for this and so for a while I had some

friends that looked at Kerberos 5 and for a while they were saying oh yeah there's still you know people out there they're using Kerberos 5 with the DES option and so I thought I'd just take a look at it and it had been known for a while that you can do downgrade attacks on Kerberos and and so that basically kind of affects like the whole you know all the ticket issuing and stuff like that and and there's there's plenty of known plaintext throughout all the communication that goes across I'll be getting into in a second so what I did was I created an editor cap filter that basically just modifies the supported encryption types to all be does CBC CRC

so like over here you can actually just see does CBC CRC there's like six times that it says oh we support des and and then this actually works for for most clients that that you know do support des so forces them to use that and then because everything's encoded in ace and one there's lots of static fields in there that are you know pretty much always there just because that's just part of the protocol and so there's plenty of known plaintext so we can use to you know derive enough information to crack the key and because CBC just relies on with the block chaining it just rides on the ciphertext of the previous block you don't really you can

basically take any block within that and just take the ciphertext of the previous block and then your known plaintext and and all of that and then be able to crack any individual block within the chain so basically makes it fairly easy for us to crack as long as you can find some plaintext and so I put together a little tool that you just feed a pcap file of the Kerberos authentication and after you do the downgrade attack or if they for some reason only use des in this you know you don't have to do that and then it'll actually pull out all the known plaintext create submission tokens you can feed that into cloud cracker or

to crack that Sh and and we'll send you the actual des key that that they that they're using for the authentication so the only issue with this is that I have no idea what to do with that des key now I've cracked a bunch of these but I need there needs to be software written to actually you know authenticate to hosts and or you know support built into in packet or Kerberos turd some of these other tools out there to actually make use of this but but at least you know everybody should know this is easy to to break those keys right now and in my day job I actually just double-checked the support of

encryption types for network and and we totally support des and so I'm I'm sure any you know a lot of a lot of environments out there that have any sort of legacy hardware on the network will probably support it as well so yeah then you can pop it on the website for these because it takes a little longer time it's thirty dollars to crack a key but usually finishes within 24 hours so uh then after having this the site up for awhile I started getting these weird emails asking if I can crack des crypt does anybody remember des crypt from a long time ago like with old UNIX systems and stuff like that is 25 rounds of des

and I think the max number of characters you type in is like 8 and just truncates the rest kind of similar to the talk that was going on before the break so so this this algorithm was originally designed so that a pdp-11 would take more than one second to compute the hash so they're like oh this will prevent everybody from from cracking this and so you think that like something like this where I think that they started using this in the seventh early 70s mid 70s nobody would be using this anymore right so it turns out that Q and X they support either des crypt the 25 round version or their own proprietary one that's fully reversible so des crypt is

actually their most secure version of a password hash that you can use and and there's at least from this press release from over two years ago there's over 50 million vehicles with Q and X according to Q and X and this password hash listed in here is actually out of the Charlie Miller and Chris valasek paper and when they hacked the jeeps and so I just like start googling around trying to find random you know des crypt passwords I found this other presentation online there's all these like forums where people just post random hashes and there's actually quite a few out there I was very surprised and so I was like okay this is great I I can like you know code up

something that'll try every possible password for this because you know it's definitely within the realm of brute-forcing and we can just add that to the service and it'll be great all these they'll just be totally totally broken so if you look at all the possible title characters you know it takes around three days to to brute-force this with the system that I had so I'm like okay I got this one from the Charlie Miller and Chris valasek paper I'm gonna crack that one so I throw it in there and I crack it and it must be super secure right all these must be super secure I'm sure I'm sure and so I even got this one from Karl

Koecher that he reversed engineered some like I think it's on star systems out there and you got the shadow file from it and so that one is just route route the one in the in the jeep packing paper is DT donkey the one and that in the Southard presentation I found is just a you know eight characters a random characters and so like if anybody here has any of these all des crypt hashes that you haven't been able to crack we're just normal john the ripper please let me know because i'd really like to find a good use for this service but but instead i've wasted you know three days of a ridiculously expensive fpga system

to crack really simple passwords so so if you if you have any ideas of things that still used as we have you know simple API that you can use to to create your own tokens to submit to the system and you can verified your implementation there's a you know it's a lot of Python so it's easy to understand and create your own tokens and then you just drop it in and you get a nice email with a you know text file with all the keys that that match the criteria that you gave it so after you put this together it's still you know it seemed like there still wasn't much being done to phase a

lot of these things out and fix the problem and it seemed to us the biggest problem with the service was that we still you know we chart who have to charge money for it and really that's just a factor of rate limiting like we could totally offer this for free but we don't want you know a backlog of you know a couple years because it does take you know a full day to crack these and you know power costs money it's expensive running you know running a system that you know drives a couple kilowatts and normally costs you know a couple hundred thousand dollars so we ask ourselves the question what if we could make this free how can we make the

service free for everybody to use and and still you know not have backlogs of a year or two so especially after seeing the stuff people are doing with cracking windows authentication started looking into creating rainbow tables for for deads and this is a pretty large key space if anybody's familiar with like the different rainbow tables out there there's like the off cracked table that does you know like a 95 ^ 8 it's they charge money for it but it's you know 2 terabytes 99% success rate I have no idea how long it takes to crack a key with it but it's pretty decent size the table that we wanted to build was run ten times bigger than that one because

it's for two of the 56 instead of around to the 52 and a half and we want to build this with 99% success rate because you want to basically make it so that if our table can't find the key then we'll just have it brute-force it and then you know save save the result of it so we can kind of you know over time be able to you know kind of fill on the gaps the table isn't able to fill and and we wanted to make it so the speed of cracking was fast enough that it wouldn't take a year if we got a backlog of jobs so looking at the hardware that we wanted to use

for this the nvme drives that are out nowadays they're actually really amazing does anybody has anybody looked at the new nvme drives a couple people so as far as doing rainbow tables really a big factor is your lookup speed and a lot of them you need drives nowadays especially if you're doing really short reads can get up to like a million I ops per second and they only advertise about half a million but but especially with really short reads they're they're really ridiculous and and for a really fast storage like this six terabytes is somewhere around you know maybe eight nine hundred dollars per per terabyte and and so you know for around five thousand dollars we could have something

that's pretty screamin fast and then we're able to borrow borrow some FPGAs for the system and it seemed like twelve would be a reasonable number to borrow and then we also borrowed another server from the server room I don't think they'll miss it and and so with it with those those parameters we came up with some some table parameters roughly half a million links in a chain and then around you know a quarter of a trillion chains per table and each table would be on two terabytes and three of those to fill up the six terabytes and that would give us around like less than three seconds and I think around like ninety-eight percent or something like

that so pretty close and super fast so we borrowed some hardware and we just happened to have some hardware that wasn't sold yet so we're like oh we'll just burn in this hardware see if it see if it works and so I filled up a system with with 48 FPGAs that we normally sell for I think like four or five thousand dollars apiece or something like that and get her done but then there were some issues we especially cramming that money in there there was some overheating problems I don't think that the previous engineers had ever run something this this intense on these FPGAs another another issue with hardware especially with stuff like this is that

power supplies are normally designed to go from say like you know a couple amps up to 50 or 60 amps like like that and so we ended up having an FPGA is just dropping out just you know even just like loading the FPGA or starting off a couple jobs so we had to put in a bunch of delays in it so it's slowly phased in you know which cores were being activated so it would be a nice smooth transition with the with the stepping and and then even on top of that like we ended up over current in the supplies I think they had like 40 amp supplies on them and we were drying a bit more than

that so so we had some hardware issues but you know took a while to get over that we got over it and so now we're okay okay let's get going we got got this fixed and spent weeks generating tables and then I don't know if anybody here has generated rainbow tables before but we spent weeks generating them and then went to sort them and then there was just like a really high collision rate and so that basically meant that the the tables be generated over the few weeks were basically unusable so learn some lessons and then I then came up with some new parameters and so we basically just had to essentially make the attack a little slower and so trade

off is basically that now we have less chains per table which means that we have more tables then it requires that much more time to actually compute your chains forward and stuff like that for doing the real-time phase but one one advantage is it because we have a better collision rate we have a lot better coverage so now we should be a lot closer to hundred percent so then we try again and the only problem is that now some of the borrowed hardware it had to go out to customers so now we're dealing with a lot less hardware and otherwise good to go fast for a few weeks but that was actually a week ago so long story short

it's we're supposed to have this fully done before now but right now we're only about halfway there but it is up so the cracking system is up this is another system that I borrowed and we put in some nice packing foam to redirect err on the FPGA is these are all the nvme drives and and so systems up and running and we only have 50% of the tables generate right now but with rainbow tables that actually means for over 90% we're at 92 percent right now which is actually pretty decent and and it's free which is great right and we're getting closer every day and actually will be at 99.5% next week so pretty much yet right after I get

back from DEFCON we'll be up to close to hundred percent and and this takes because we're cracking two keys it takes around 24 seconds or something like that per job so we have time for a quick demo oh yeah we got plenty of time so I just wanted a demo using responder to crack a net ntlm hash to an NT hash and then see what we can do with that so oh yeah let me get up my notes real quick I always mess up demos if I don't have my notes okay so we have here is Kali Linux and Windows 7 so I'm just gonna both these are on the same internal network it's gonna fire

per sponder with the land man downgrade option so that's fired up and no yeah I mean obviously in the production environment we're gonna be browsing to real shares but so I just put in blood respite response to it and now I have an ntlm v1 hash right here and so normally we take something like this and put it into John the Ripper but we're gonna take that net ntlm hash and another thing of note two is that in responder I set the challenge to the 1 1 2 2 3 3 4 4 and that's important for the rainbow table to work I hope it goes it's only built for that specific plaintext so now that I have

that hash I'm going to switch over to cracked-out SH and as long as nobody killed my internet connection here I put t hash so now it just asked for an email there's no payment or anything I put in my email address submit and pray to the demo gods maybe try something one more time yeah that's working somebody just decided this would be a great time to DDoS crack that Sh all right well yeah I mean this was put together basically last night around 3:00 a.m. so so that'll definitely be going up soon well so worst case I have a bunch of ones in my trash here let's see er no it's the last one I did

okay so this is my you basically get an email kind of like this with the NT hash so no no we can go over to here and so now with the NT hash I'll say let's do this and B client so I paste our hash I know okay so we're on the system that's pretty cool and then if I miss Floyd or if you have like a PSX Zack or SMB exec or something like that I'm just gonna create a shell over to it just so we can see that

so we're gonna do a oh oh wait and we trigger sorry all right and then uh so then we have it call back to us set our Samba password which is gonna be our hash user no okay so now connected and ferpa shell oops and we're a system so no yeah you can do whatever you want on the system using then T hash so and there's there's tons of other tools out there for using NT hashes - you know exploit systems and stuff so so I mean that's that's the basic demo and see if this and they've actually gone through oh yeah so yeah when it actually goes through then it'll give you a little page we can check your

status and then it'll send you an email and then we also have a API that we're we're releasing here at Def Con we're gonna be in the demo labs room if you want to find us over there we'll be demoing a few different things and so we're setting up a rest interface we'll have to figure out a way to prevent people from abusing that too but for now we actually had somebody contact us recently and they put together an air bot check plugins so you can submit stuff through slack or IRC or whatever I don't know what use chat BOTS for and and then we're also gonna possibly integrate with like hostapd wpe so it can just automatically you know

crack the hashes or we could even integrate directly with responders so you know as it finds those and it automatically just reverses them to NT hashes and displays that so there's lots of fun that we could do with this and so if anybody's interested in writing some stuff using our API let me know and I'll get you some more info on it and then of course there's lots of people that have kind of helped out with all the related research so I just thought I'd include them in there and otherwise help me kill the legacy crypto I'm calling on everyone here if you have any ideas or you know want to help out with this at all let me know and if you

want to run me free jobs just email me I mean it really is just for rate-limiting the system so shoot me an email if you have any ideas everything is on github cracked out SH and I also helped run toric on which is going on next month the main conference is September 1st through the 3rd so I hope to see some of you there and then we also throw a really amazing event called tour camp that's happening next year and it's the the us hacker camp we're out on beautiful island up in the northwest and so that's always a lot of fun thank you very much

any questions oh sorry what's that oh yeah ok so would you consider triple des legacy that should be yeah I mean I think most people would consider a legacy but yeah it's nowhere near I mean it's not something that I can crack with the hardware that I have laying around at my company I'm sure you know like there ya with enough storage and you know dedicated hardware I'm sure you know governments and well-funded organizations could do that but but yeah it's definitely well out of you know my parameters of like a real attack is something you could do within 24 hours or you know within seconds so yeah so how long did it take to compute the

table and any idea how long it would take to do that on f1 well so yeah so I probably should include some f1 numbers um so with with the tables I mean it's kind of it's been a hot podge of like generating tables and unlike various systems around the office I think the total amount of time for each table just just generating is around 48 hours with like 12 FPGAs and then and then we have to do 12 of those and so I don't know it depends on how many FPGAs you have and stuff like that and I'd have to work out the math on how equivalent that is to the I think it's like the the v9

P's that that f1 has but yeah I mean I imagine that it wouldn't it wouldn't take that long to generate something like that on f1 that I would have done it on f1 except I was mostly concerned about the like the speed of the drives for doing the the real time attack and stuff like that and so I guess I need to I need to just see how how fast the drives are if you know we could do be doing close to like a million you know I ops per second with with something like that and then another another consideration too is that I looked at switching over this whole service to f1 and just an f1 credits it would cost

more than the twenty dollars a month or whatever that we chart or twenty dollars per key that we're charging and then especially as a free service and that's something that we have to pay you know and so it's a lot easier when we can just kind of borrow hardware and you know but yeah yeah yeah yeah I mean it you know it probably would have worked pretty well fruit for generating the tables just as far as scaling up and especially with the the sorting process takes a lot of time to but then yeah once you have the tables and it's a matter of you know getting the six terabytes out of out of the UAC to or

the storage and so so yeah I don't know it seemed at least for me it was easier to do it do it on our hardware but yeah it seems like it probably wouldn't take that long on f1 hardware it mostly depends on how much money you have to spend are you looking to make those tables available yeah sure yeah I I need to clean up all the code and stuff like that and figure out a good way of distributing six terabytes but yeah definitely if there's people are willing to torrent it or whatever then Danielle I'll figure out a way to get it out there and I have no idea how long it would take on data duplication village

to help you know yeah there we go yeah yeah somebody'll bring some hard drives next time or something and yeah there's still needs to be selling the right support for you know running on GPUs or CPUs or other sorts of hardware to so I was very specific for for FPGAs and nvme drives and all that any other questions oh oh oh just sunny normal like magnetic storage sure yeah you know if you if you copy those things onto spinning hard you know regular spinning commercial hard drives yeah yeah I mean I I don't know what modern like 7200 rpm drives are I thought it's in like the tens of thousands of AI ops or something like

that so yeah I mean I if you're running on a CPU would probably be totally fine because everything's gonna be a lot slower anyway but if you're really willing to wait you know hours or days then it's yeah it probably doesn't matter what storage you're keeping it on

which versions of Windows still support beds out of the gate Oh for like Kerberos surfer for like the landmasses downgrade attack which versions of Windows can't pull it off within which one's left to pray for umm well so by default I think it I think it's probably like you know Windows 2000 and maybe maybe Windows 2003 or at least like we pre-service Packard and XP and stuff like that yeah definitely it's definitely fixed in Vista and newer but but yeah like for our environment you know we have Windows 7 and Windows 10 and stuff like that on our corporate network and all of them only support our c4 in DES just because we have tons of

other systems out there and and I guess also there's there's a number of systems like for handling like all of your you know IT account creation and you know password maintenance and stuff like that that they basically reverse engineered all these protocols so they do only support the really old ones like our c4 and DES to be able to do all that stuff on the network so see I'd really check your and your supported encryption types on your network to make sure because yeah just because you're running Windows 10 doesn't mean that you don't have DES as an option

if do you mind a UNIX MIT Kerberos question okay okay so like say you've upgraded your host keys and your user keys to Triple DES or AES uh-huh the at least historically the Kerberos master database password mm-hmm has you initially set up DES and it's been difficult to change oh but I think they said it's okay cause it's not really exposed huh and any ideas about that yeah I mean I when I was when I was setting up this whole thing to to the whole TAC environment and stuff I'd I I set up just for like the standard kirb kirb five that comes with the bun too and and it wouldn't let me have my

master key beads single des I think and so I don't know how legacy you have to go back before you know they actually let you do that but but yeah I I'm not that I'm honestly not that familiar with Kerberos so just yeah yeah but yeah I'm sure a lot a lot of the older systems where des was a lot more prevalent they'd yeah be a lot pretty difficult to switch from that alright any other questions all right well I'll be around after the talks ode yeah just um come on up if you have any more questions thanks a lot [Applause]